bibtex_url
stringlengths 41
50
| proceedings
stringlengths 38
47
| bibtext
stringlengths 709
3.56k
| abstract
stringlengths 17
2.11k
| authors
sequencelengths 1
72
| title
stringlengths 12
207
| id
stringlengths 7
16
| type
stringclasses 2
values | arxiv_id
stringlengths 0
10
| GitHub
sequencelengths 1
1
| paper_page
stringclasses 276
values | n_linked_authors
int64 -1
13
| upvotes
int64 -1
14
| num_comments
int64 -1
11
| n_authors
int64 -1
44
| paper_page_exists_pre_conf
int64 0
1
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
14
| Spaces
sequencelengths 0
100
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://aclanthology.org/2023.acl-long.500.bib | https://aclanthology.org/2023.acl-long.500/ | @inproceedings{zeng-etal-2023-synthesize,
title = "Synthesize, Prompt and Transfer: Zero-shot Conversational Question Generation with Pre-trained Language Model",
author = "Zeng, Hongwei and
Wei, Bifan and
Liu, Jun and
Fu, Weiping",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.500",
doi = "10.18653/v1/2023.acl-long.500",
pages = "8989--9010",
abstract = "Conversational question generation aims to generate questions that depend on both context and conversation history. Conventional works utilizing deep learning have shown promising results, but heavily rely on the availability of large-scale annotated conversations. In this paper, we introduce a more realistic and less explored setting, Zero-shot Conversational Question Generation (ZeroCQG), which requires no human-labeled conversations for training. To solve ZeroCQG, we propose a multi-stage knowledge transfer framework, Synthesize, Prompt, and trAnsfer with pRe-Trained lAnguage model (SPARTA) to effectively leverage knowledge from single-turn question generation instances. To validate the zero-shot performance of SPARTA, we conduct extensive experiments on three conversational datasets: CoQA, QuAC, and DoQA by transferring knowledge from three single-turn datasets: MS MARCO, NewsQA, and SQuAD. The experimental results demonstrate the superior performance of our method. Specifically, SPARTA has achieved 14.81 BLEU-4 (88.2{\%} absolute improvement compared to T5) in CoQA with knowledge transferred from SQuAD.",
}
| Conversational question generation aims to generate questions that depend on both context and conversation history. Conventional works utilizing deep learning have shown promising results, but heavily rely on the availability of large-scale annotated conversations. In this paper, we introduce a more realistic and less explored setting, Zero-shot Conversational Question Generation (ZeroCQG), which requires no human-labeled conversations for training. To solve ZeroCQG, we propose a multi-stage knowledge transfer framework, Synthesize, Prompt, and trAnsfer with pRe-Trained lAnguage model (SPARTA) to effectively leverage knowledge from single-turn question generation instances. To validate the zero-shot performance of SPARTA, we conduct extensive experiments on three conversational datasets: CoQA, QuAC, and DoQA by transferring knowledge from three single-turn datasets: MS MARCO, NewsQA, and SQuAD. The experimental results demonstrate the superior performance of our method. Specifically, SPARTA has achieved 14.81 BLEU-4 (88.2{\%} absolute improvement compared to T5) in CoQA with knowledge transferred from SQuAD. | [
"Zeng, Hongwei",
"Wei, Bifan",
"Liu, Jun",
"Fu, Weiping"
] | Synthesize, Prompt and Transfer: Zero-shot Conversational Question Generation with Pre-trained Language Model | acl-long.500 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.501.bib | https://aclanthology.org/2023.acl-long.501/ | @inproceedings{lee-etal-2023-formnetv2,
title = "{F}orm{N}et{V}2: Multimodal Graph Contrastive Learning for Form Document Information Extraction",
author = "Lee, Chen-Yu and
Li, Chun-Liang and
Zhang, Hao and
Dozat, Timothy and
Perot, Vincent and
Su, Guolong and
Zhang, Xiang and
Sohn, Kihyuk and
Glushnev, Nikolay and
Wang, Renshen and
Ainslie, Joshua and
Long, Shangbang and
Qin, Siyang and
Fujii, Yasuhisa and
Hua, Nan and
Pfister, Tomas",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.501",
doi = "10.18653/v1/2023.acl-long.501",
pages = "9011--9026",
abstract = "The recent advent of self-supervised pre-training techniques has led to a surge in the use of multimodal learning in form document understanding. However, existing approaches that extend the mask language modeling to other modalities require careful multi-task tuning, complex reconstruction target designs, or additional pre-training data. In FormNetV2, we introduce a centralized multimodal graph contrastive learning strategy to unify self-supervised pre-training for all modalities in one loss. The graph contrastive objective maximizes the agreement of multimodal representations, providing a natural interplay for all modalities without special customization. In addition, we extract image features within the bounding box that joins a pair of tokens connected by a graph edge, capturing more targeted visual cues without loading a sophisticated and separately pre-trained image embedder. FormNetV2 establishes new state-of-the-art performance on FUNSD, CORD, SROIE and Payment benchmarks with a more compact model size.",
}
| The recent advent of self-supervised pre-training techniques has led to a surge in the use of multimodal learning in form document understanding. However, existing approaches that extend the mask language modeling to other modalities require careful multi-task tuning, complex reconstruction target designs, or additional pre-training data. In FormNetV2, we introduce a centralized multimodal graph contrastive learning strategy to unify self-supervised pre-training for all modalities in one loss. The graph contrastive objective maximizes the agreement of multimodal representations, providing a natural interplay for all modalities without special customization. In addition, we extract image features within the bounding box that joins a pair of tokens connected by a graph edge, capturing more targeted visual cues without loading a sophisticated and separately pre-trained image embedder. FormNetV2 establishes new state-of-the-art performance on FUNSD, CORD, SROIE and Payment benchmarks with a more compact model size. | [
"Lee, Chen-Yu",
"Li, Chun-Liang",
"Zhang, Hao",
"Dozat, Timothy",
"Perot, Vincent",
"Su, Guolong",
"Zhang, Xiang",
"Sohn, Kihyuk",
"Glushnev, Nikolay",
"Wang, Renshen",
"Ainslie, Joshua",
"Long, Shangbang",
"Qin, Siyang",
"Fujii, Yasuhisa",
"Hua, Nan",
"Pfister, Tomas"
] | FormNetV2: Multimodal Graph Contrastive Learning for Form Document Information Extraction | acl-long.501 | Poster | 2305.02549 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.502.bib | https://aclanthology.org/2023.acl-long.502/ | @inproceedings{zhang-etal-2023-mixce,
title = "{M}ix{CE}: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies",
author = "Zhang, Shiyue and
Wu, Shijie and
Irsoy, Ozan and
Lu, Steven and
Bansal, Mohit and
Dredze, Mark and
Rosenberg, David",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.502",
doi = "10.18653/v1/2023.acl-long.502",
pages = "9027--9050",
abstract = "Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P {--} that is, minimizing the forward cross-entropy, which is equivalent to maximum likelihood estimation (MLE). We have observed that models trained in this way may {``}over-generalize{''}, in the sense that they produce non-human-like text. Moreover, we believe that reverse cross-entropy, i.e., the cross-entropy of P relative to Q, is a better reflection of how a human would evaluate text generated by a model. Hence, we propose learning with MixCE, an objective that mixes the forward and reverse cross-entropies. We evaluate models trained with this objective on synthetic data settings (where P is known) and real data, and show that the resulting models yield better generated text without complex decoding strategies.",
}
| Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P {--} that is, minimizing the forward cross-entropy, which is equivalent to maximum likelihood estimation (MLE). We have observed that models trained in this way may {``}over-generalize{''}, in the sense that they produce non-human-like text. Moreover, we believe that reverse cross-entropy, i.e., the cross-entropy of P relative to Q, is a better reflection of how a human would evaluate text generated by a model. Hence, we propose learning with MixCE, an objective that mixes the forward and reverse cross-entropies. We evaluate models trained with this objective on synthetic data settings (where P is known) and real data, and show that the resulting models yield better generated text without complex decoding strategies. | [
"Zhang, Shiyue",
"Wu, Shijie",
"Irsoy, Ozan",
"Lu, Steven",
"Bansal, Mohit",
"Dredze, Mark",
"Rosenberg, David"
] | MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies | acl-long.502 | Poster | 2305.16958 | [
"https://github.com/bloomberg/mixce-acl2023"
] | https://huggingface.co/papers/2305.16958 | 2 | 1 | 0 | 7 | 1 | [
"euclaise/Memphis-CoT-3B"
] | [] | [] |
https://aclanthology.org/2023.acl-long.503.bib | https://aclanthology.org/2023.acl-long.503/ | @inproceedings{zhao-etal-2023-knowledgeable,
title = "Knowledgeable Parameter Efficient Tuning Network for Commonsense Question Answering",
author = "Zhao, Ziwang and
Hu, Linmei and
Zhao, Hanyu and
Shao, Yingxia and
Wang, Yequan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.503",
doi = "10.18653/v1/2023.acl-long.503",
pages = "9051--9063",
abstract = "Commonsense question answering is important for making decisions about everyday matters. Although existing commonsense question answering works based on fully fine-tuned PLMs have achieved promising results, they suffer from prohibitive computation costs as well as poor interpretability. Some works improve the PLMs by incorporating knowledge to provide certain evidence, via elaborately designed GNN modules which require expertise. In this paper, we propose a simple knowledgeable parameter efficient tuning network to couple PLMs with external knowledge for commonsense question answering. Specifically, we design a trainable parameter-sharing adapter attached to a parameter-freezing PLM to incorporate knowledge at a small cost. The adapter is equipped with both entity- and query-related knowledge via two auxiliary knowledge-related tasks (i.e., span masking and relation discrimination). To make the adapter focus on the relevant knowledge, we design gating and attention mechanisms to respectively filter and fuse the query information from the PLM. Extensive experiments on two benchmark datasets show that KPE is parameter-efficient and can effectively incorporate knowledge for improving commonsense question answering.",
}
| Commonsense question answering is important for making decisions about everyday matters. Although existing commonsense question answering works based on fully fine-tuned PLMs have achieved promising results, they suffer from prohibitive computation costs as well as poor interpretability. Some works improve the PLMs by incorporating knowledge to provide certain evidence, via elaborately designed GNN modules which require expertise. In this paper, we propose a simple knowledgeable parameter efficient tuning network to couple PLMs with external knowledge for commonsense question answering. Specifically, we design a trainable parameter-sharing adapter attached to a parameter-freezing PLM to incorporate knowledge at a small cost. The adapter is equipped with both entity- and query-related knowledge via two auxiliary knowledge-related tasks (i.e., span masking and relation discrimination). To make the adapter focus on the relevant knowledge, we design gating and attention mechanisms to respectively filter and fuse the query information from the PLM. Extensive experiments on two benchmark datasets show that KPE is parameter-efficient and can effectively incorporate knowledge for improving commonsense question answering. | [
"Zhao, Ziwang",
"Hu, Linmei",
"Zhao, Hanyu",
"Shao, Yingxia",
"Wang, Yequan"
] | Knowledgeable Parameter Efficient Tuning Network for Commonsense Question Answering | acl-long.503 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.504.bib | https://aclanthology.org/2023.acl-long.504/ | @inproceedings{chen-etal-2023-blaser,
title = "{BLASER}: A Text-Free Speech-to-Speech Translation Evaluation Metric",
author = "Chen, Mingda and
Duquenne, Paul-Ambroise and
Andrews, Pierre and
Kao, Justine and
Mourachko, Alexandre and
Schwenk, Holger and
Costa-juss{\`a}, Marta R.",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.504",
doi = "10.18653/v1/2023.acl-long.504",
pages = "9064--9079",
abstract = "End-to-End speech-to-speech translation (S2ST) is generally evaluated with text-based metrics. This means that generated speech has to be automatically transcribed, making the evaluation dependent on the availability and quality of automatic speech recognition (ASR) systems. In this paper, we propose a text-free evaluation metric for end-to-end S2ST, named BLASER, to avoid the dependency on ASR systems. BLASER leverages a multilingual multimodal encoder to directly encode the speech segments for source input, translation output and reference into a shared embedding space and computes a score of the translation quality that can be used as a proxy to human evaluation. To evaluate our approach, we construct training and evaluation sets from more than 40k human annotations covering seven language directions. The best results of BLASER are achieved by training with supervision from human rating scores. We show that when evaluated at the sentence level, BLASER correlates significantly better with human judgment compared to ASR dependent metrics including ASR-SENTBLEU in all translation directions and ASR-COMET in five of them. Our analysis shows combining speech and text as inputs to BLASER does not increase the correlation with human scores, but best correlations are achieved when using speech, which motivates the goal of our research. Moreover, we show that using ASR for references is detrimental for text-based metrics.",
}
| End-to-End speech-to-speech translation (S2ST) is generally evaluated with text-based metrics. This means that generated speech has to be automatically transcribed, making the evaluation dependent on the availability and quality of automatic speech recognition (ASR) systems. In this paper, we propose a text-free evaluation metric for end-to-end S2ST, named BLASER, to avoid the dependency on ASR systems. BLASER leverages a multilingual multimodal encoder to directly encode the speech segments for source input, translation output and reference into a shared embedding space and computes a score of the translation quality that can be used as a proxy to human evaluation. To evaluate our approach, we construct training and evaluation sets from more than 40k human annotations covering seven language directions. The best results of BLASER are achieved by training with supervision from human rating scores. We show that when evaluated at the sentence level, BLASER correlates significantly better with human judgment compared to ASR dependent metrics including ASR-SENTBLEU in all translation directions and ASR-COMET in five of them. Our analysis shows combining speech and text as inputs to BLASER does not increase the correlation with human scores, but best correlations are achieved when using speech, which motivates the goal of our research. Moreover, we show that using ASR for references is detrimental for text-based metrics. | [
"Chen, Mingda",
"Duquenne, Paul-Ambroise",
"Andrews, Pierre",
"Kao, Justine",
"Mourachko, Alex",
"re",
"Schwenk, Holger",
"Costa-juss{\\`a}, Marta R."
] | BLASER: A Text-Free Speech-to-Speech Translation Evaluation Metric | acl-long.504 | Poster | 2212.08486 | [
"https://github.com/facebookresearch/stopes"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.505.bib | https://aclanthology.org/2023.acl-long.505/ | @inproceedings{santy-etal-2023-nlpositionality,
title = "{NLP}ositionality: Characterizing Design Biases of Datasets and Models",
author = "Santy, Sebastin and
Liang, Jenny and
Le Bras, Ronan and
Reinecke, Katharina and
Sap, Maarten",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.505",
doi = "10.18653/v1/2023.acl-long.505",
pages = "9080--9102",
abstract = "Design biases in NLP systems, such as performance differences for different populations, often stem from their creator{'}s positionality, i.e., views and lived experiences shaped by identity and background. Despite the prevalence and risks of design biases, they are hard to quantify because researcher, system, and dataset positionality is often unobserved. We introduce NLPositionality, a framework for characterizing design biases and quantifying the positionality of NLP datasets and models. Our framework continuously collects annotations from a diverse pool of volunteer participants on LabintheWild, and statistically quantifies alignment with dataset labels and model predictions. We apply NLPositionality to existing datasets and models for two tasks{---}social acceptability and hate speech detection. To date, we have collected 16,299 annotations in over a year for 600 instances from 1,096 annotators across 87 countries. We find that datasets and models align predominantly with Western, White, college-educated, and younger populations. Additionally, certain groups, such as non-binary people and non-native English speakers, are further marginalized by datasets and models as they rank least in alignment across all tasks. Finally, we draw from prior literature to discuss how researchers can examine their own positionality and that of their datasets and models, opening the door for more inclusive NLP systems.",
}
| Design biases in NLP systems, such as performance differences for different populations, often stem from their creator{'}s positionality, i.e., views and lived experiences shaped by identity and background. Despite the prevalence and risks of design biases, they are hard to quantify because researcher, system, and dataset positionality is often unobserved. We introduce NLPositionality, a framework for characterizing design biases and quantifying the positionality of NLP datasets and models. Our framework continuously collects annotations from a diverse pool of volunteer participants on LabintheWild, and statistically quantifies alignment with dataset labels and model predictions. We apply NLPositionality to existing datasets and models for two tasks{---}social acceptability and hate speech detection. To date, we have collected 16,299 annotations in over a year for 600 instances from 1,096 annotators across 87 countries. We find that datasets and models align predominantly with Western, White, college-educated, and younger populations. Additionally, certain groups, such as non-binary people and non-native English speakers, are further marginalized by datasets and models as they rank least in alignment across all tasks. Finally, we draw from prior literature to discuss how researchers can examine their own positionality and that of their datasets and models, opening the door for more inclusive NLP systems. | [
"Santy, Sebastin",
"Liang, Jenny",
"Le Bras, Ronan",
"Reinecke, Katharina",
"Sap, Maarten"
] | NLPositionality: Characterizing Design Biases of Datasets and Models | acl-long.505 | Oral | 2306.01943 | [
"https://github.com/liang-jenny/nlpositionality"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.506.bib | https://aclanthology.org/2023.acl-long.506/ | @inproceedings{hewitt-etal-2023-backpack,
title = "Backpack Language Models",
author = "Hewitt, John and
Thickstun, John and
Manning, Christopher and
Liang, Percy",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.506",
doi = "10.18653/v1/2023.acl-long.506",
pages = "9103--9125",
abstract = "We present Backpacks: a new neural architecture that marries strong modeling performancewith an interface for interpretability and control. Backpacks learn multiple non-contextual sense vectors for each word in a vocabulary, and represent a word in a sequence as a context-dependent, non-negative linear combination ofsense vectors in this sequence. We find that, after training, sense vectors specialize, each encoding a different aspect of a word. We can interpret a sense vector by inspecting its (non-contextual, linear) projection onto the output space, and intervene on these interpretable hooks to change the model{'}s behavior in predictable ways. We train a 170M-parameter Backpack language model on OpenWebText, matching the loss of a GPT-2 small (124Mparameter) Transformer. On lexical similarity evaluations, we find that Backpack sense vectors outperform even a 6B-parameter Transformer LM{'}s word embeddings. Finally, we present simple algorithms that intervene on sense vectors to perform controllable text generation and debiasing. For example, we can edit the sense vocabulary to tend more towards a topic, or localize a source of gender bias to a sense vector and globally suppress that sense.",
}
| We present Backpacks: a new neural architecture that marries strong modeling performancewith an interface for interpretability and control. Backpacks learn multiple non-contextual sense vectors for each word in a vocabulary, and represent a word in a sequence as a context-dependent, non-negative linear combination ofsense vectors in this sequence. We find that, after training, sense vectors specialize, each encoding a different aspect of a word. We can interpret a sense vector by inspecting its (non-contextual, linear) projection onto the output space, and intervene on these interpretable hooks to change the model{'}s behavior in predictable ways. We train a 170M-parameter Backpack language model on OpenWebText, matching the loss of a GPT-2 small (124Mparameter) Transformer. On lexical similarity evaluations, we find that Backpack sense vectors outperform even a 6B-parameter Transformer LM{'}s word embeddings. Finally, we present simple algorithms that intervene on sense vectors to perform controllable text generation and debiasing. For example, we can edit the sense vocabulary to tend more towards a topic, or localize a source of gender bias to a sense vector and globally suppress that sense. | [
"Hewitt, John",
"Thickstun, John",
"Manning, Christopher",
"Liang, Percy"
] | Backpack Language Models | acl-long.506 | Poster | 2305.16765 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.507.bib | https://aclanthology.org/2023.acl-long.507/ | @inproceedings{felkner-etal-2023-winoqueer,
title = "{W}ino{Q}ueer: A Community-in-the-Loop Benchmark for Anti-{LGBTQ}+ Bias in Large Language Models",
author = "Felkner, Virginia and
Chang, Ho-Chun Herbert and
Jang, Eugene and
May, Jonathan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.507",
doi = "10.18653/v1/2023.acl-long.507",
pages = "9126--9140",
abstract = "We present WinoQueer: a benchmark specifically designed to measure whether large language models (LLMs) encode biases that are harmful to the LGBTQ+ community. The benchmark is community-sourced, via application of a novel method that generates a bias benchmark from a community survey. We apply our benchmark to several popular LLMs and find that off-the-shelf models generally do exhibit considerable anti-queer bias. Finally, we show that LLM bias against a marginalized community can be somewhat mitigated by finetuning on data written about or by members of that community, and that social media text written by community members is more effective than news text written about the community by non-members. Our method for community-in-the-loop benchmark development provides a blueprint for future researchers to develop community-driven, harms-grounded LLM benchmarks for other marginalized communities.",
}
| We present WinoQueer: a benchmark specifically designed to measure whether large language models (LLMs) encode biases that are harmful to the LGBTQ+ community. The benchmark is community-sourced, via application of a novel method that generates a bias benchmark from a community survey. We apply our benchmark to several popular LLMs and find that off-the-shelf models generally do exhibit considerable anti-queer bias. Finally, we show that LLM bias against a marginalized community can be somewhat mitigated by finetuning on data written about or by members of that community, and that social media text written by community members is more effective than news text written about the community by non-members. Our method for community-in-the-loop benchmark development provides a blueprint for future researchers to develop community-driven, harms-grounded LLM benchmarks for other marginalized communities. | [
"Felkner, Virginia",
"Chang, Ho-Chun Herbert",
"Jang, Eugene",
"May, Jonathan"
] | WinoQueer: A Community-in-the-Loop Benchmark for Anti-LGBTQ+ Bias in Large Language Models | acl-long.507 | Oral | 2306.15087 | [
"https://github.com/katyfelkner/winoqueer"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.508.bib | https://aclanthology.org/2023.acl-long.508/ | @inproceedings{yu-etal-2023-grounded,
title = "Grounded Multimodal Named Entity Recognition on Social Media",
author = "Yu, Jianfei and
Li, Ziyan and
Wang, Jieming and
Xia, Rui",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.508",
doi = "10.18653/v1/2023.acl-long.508",
pages = "9141--9154",
abstract = "In recent years, Multimodal Named Entity Recognition (MNER) on social media has attracted considerable attention. However, existing MNER studies only extract entity-type pairs in text, which is useless for multimodal knowledge graph construction and insufficient for entity disambiguation. To solve these issues, in this work, we introduce a Grounded Multimodal Named Entity Recognition (GMNER) task. Given a text-image social post, GMNER aims to identify the named entities in text, their entity types, and their bounding box groundings in image (i.e. visual regions). To tackle the GMNER task, we construct a Twitter dataset based on two existing MNER datasets. Moreover, we extend four well-known MNER methods to establish a number of baseline systems and further propose a Hierarchical Index generation framework named H-Index, which generates the entity-type-region triples in a hierarchical manner with a sequence-to-sequence model. Experiment results on our annotated dataset demonstrate the superiority of our H-Index framework over baseline systems on the GMNER task.",
}
| In recent years, Multimodal Named Entity Recognition (MNER) on social media has attracted considerable attention. However, existing MNER studies only extract entity-type pairs in text, which is useless for multimodal knowledge graph construction and insufficient for entity disambiguation. To solve these issues, in this work, we introduce a Grounded Multimodal Named Entity Recognition (GMNER) task. Given a text-image social post, GMNER aims to identify the named entities in text, their entity types, and their bounding box groundings in image (i.e. visual regions). To tackle the GMNER task, we construct a Twitter dataset based on two existing MNER datasets. Moreover, we extend four well-known MNER methods to establish a number of baseline systems and further propose a Hierarchical Index generation framework named H-Index, which generates the entity-type-region triples in a hierarchical manner with a sequence-to-sequence model. Experiment results on our annotated dataset demonstrate the superiority of our H-Index framework over baseline systems on the GMNER task. | [
"Yu, Jianfei",
"Li, Ziyan",
"Wang, Jieming",
"Xia, Rui"
] | Grounded Multimodal Named Entity Recognition on Social Media | acl-long.508 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.509.bib | https://aclanthology.org/2023.acl-long.509/ | @inproceedings{zheng-etal-2023-preserving,
title = "Preserving Commonsense Knowledge from Pre-trained Language Models via Causal Inference",
author = "Zheng, Junhao and
Ma, Qianli and
Qiu, Shengjie and
Wu, Yue and
Ma, Peitian and
Liu, Junlong and
Feng, Huawen and
Shang, Xichen and
Chen, Haibin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.509",
doi = "10.18653/v1/2023.acl-long.509",
pages = "9155--9173",
abstract = "Fine-tuning has been proven to be a simple and effective technique to transfer the learned knowledge of Pre-trained Language Models (PLMs) to downstream tasks. However, vanilla fine-tuning easily overfits the target data and degrades the generalization ability. Most existing studies attribute it to catastrophic forgetting, and they retain the pre-trained knowledge indiscriminately without identifying what knowledge is transferable. Motivated by this, we frame fine-tuning into a causal graph and discover that the crux of catastrophic forgetting lies in the missing causal effects from the pre-trained data. Based on the causal view, we propose a unified objective for fine-tuning to retrieve the causality back. Intriguingly, the unified objective can be seen as the sum of the vanilla fine-tuning objective, which learns new knowledge from target data, and the causal objective, which preserves old knowledge from PLMs. Therefore, our method is flexible and can mitigate negative transfer while preserving knowledge. Since endowing models with commonsense is a long-standing challenge, we implement our method on commonsense QA with a proposed heuristic estimation to verify its effectiveness. In the experiments, our method outperforms state-of-the-art fine-tuning methods on all six commonsense QA datasets and can be implemented as a plug-in module to inflate the performance of existing QA models.",
}
| Fine-tuning has been proven to be a simple and effective technique to transfer the learned knowledge of Pre-trained Language Models (PLMs) to downstream tasks. However, vanilla fine-tuning easily overfits the target data and degrades the generalization ability. Most existing studies attribute it to catastrophic forgetting, and they retain the pre-trained knowledge indiscriminately without identifying what knowledge is transferable. Motivated by this, we frame fine-tuning into a causal graph and discover that the crux of catastrophic forgetting lies in the missing causal effects from the pre-trained data. Based on the causal view, we propose a unified objective for fine-tuning to retrieve the causality back. Intriguingly, the unified objective can be seen as the sum of the vanilla fine-tuning objective, which learns new knowledge from target data, and the causal objective, which preserves old knowledge from PLMs. Therefore, our method is flexible and can mitigate negative transfer while preserving knowledge. Since endowing models with commonsense is a long-standing challenge, we implement our method on commonsense QA with a proposed heuristic estimation to verify its effectiveness. In the experiments, our method outperforms state-of-the-art fine-tuning methods on all six commonsense QA datasets and can be implemented as a plug-in module to inflate the performance of existing QA models. | [
"Zheng, Junhao",
"Ma, Qianli",
"Qiu, Shengjie",
"Wu, Yue",
"Ma, Peitian",
"Liu, Junlong",
"Feng, Huawen",
"Shang, Xichen",
"Chen, Haibin"
] | Preserving Commonsense Knowledge from Pre-trained Language Models via Causal Inference | acl-long.509 | Oral | 2306.10790 | [
"https://github.com/qianlima-lab/cet"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.510.bib | https://aclanthology.org/2023.acl-long.510/ | @inproceedings{li-etal-2023-translation,
title = "Translation-Enhanced Multilingual Text-to-Image Generation",
author = "Li, Yaoyiran and
Chang, Ching-Yun and
Rawls, Stephen and
Vuli{\'c}, Ivan and
Korhonen, Anna",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.510",
doi = "10.18653/v1/2023.acl-long.510",
pages = "9174--9193",
abstract = "Research on text-to-image generation (TTI) still predominantly focuses on the English language due to the lack of annotated image-caption data in other languages; in the long run, this might widen inequitable access to TTI technology. In this work, we thus investigate multilingual TTI (termed mTTI) and the current potential of neural machine translation (NMT) to bootstrap mTTI systems. We provide two key contributions. 1) Relying on a multilingual multi-modal encoder, we provide a systematic empirical study of standard methods used in cross-lingual NLP when applied to mTTI: Translate Train, Translate Test, and Zero-Shot Transfer. 2) We propose Ensemble Adapter (EnsAd), a novel parameter-efficient approach that learns to weigh and consolidate the multilingual text knowledge within the mTTI framework, mitigating the language gap and thus improving mTTI performance. Our evaluations on standard mTTI datasets COCO-CN, Multi30K Task2, and LAION-5B demonstrate the potential of translation-enhanced mTTI systems and also validate the benefits of the proposed EnsAd which derives consistent gains across all datasets. Further investigations on model variants, ablation studies, and qualitative analyses provide additional insights on the inner workings of the proposed mTTI approaches.",
}
| Research on text-to-image generation (TTI) still predominantly focuses on the English language due to the lack of annotated image-caption data in other languages; in the long run, this might widen inequitable access to TTI technology. In this work, we thus investigate multilingual TTI (termed mTTI) and the current potential of neural machine translation (NMT) to bootstrap mTTI systems. We provide two key contributions. 1) Relying on a multilingual multi-modal encoder, we provide a systematic empirical study of standard methods used in cross-lingual NLP when applied to mTTI: Translate Train, Translate Test, and Zero-Shot Transfer. 2) We propose Ensemble Adapter (EnsAd), a novel parameter-efficient approach that learns to weigh and consolidate the multilingual text knowledge within the mTTI framework, mitigating the language gap and thus improving mTTI performance. Our evaluations on standard mTTI datasets COCO-CN, Multi30K Task2, and LAION-5B demonstrate the potential of translation-enhanced mTTI systems and also validate the benefits of the proposed EnsAd which derives consistent gains across all datasets. Further investigations on model variants, ablation studies, and qualitative analyses provide additional insights on the inner workings of the proposed mTTI approaches. | [
"Li, Yaoyiran",
"Chang, Ching-Yun",
"Rawls, Stephen",
"Vuli{\\'c}, Ivan",
"Korhonen, Anna"
] | Translation-Enhanced Multilingual Text-to-Image Generation | acl-long.510 | Poster | 2305.19216 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.511.bib | https://aclanthology.org/2023.acl-long.511/ | @inproceedings{maynez-etal-2023-benchmarking,
title = "Benchmarking Large Language Model Capabilities for Conditional Generation",
author = "Maynez, Joshua and
Agrawal, Priyanka and
Gehrmann, Sebastian",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.511",
doi = "10.18653/v1/2023.acl-long.511",
pages = "9194--9213",
abstract = "Pre-trained large language models (PLMs) underly most new developments in natural language processing. They have shifted the field from application-specific model pipelines to a single model that is adapted to a wide range of tasks. Autoregressive PLMs like GPT-3 or PaLM and associated techniques like fewshot learning, have additionally shifted the output modality to generation instead of classification or regression. Despite their ubiquitous use, the generation quality of language models is rarely evaluated when these models are introduced. Additionally, it is unclear how existing generation tasks{--}while they can be used to compare systems at a high level{--}relate to the real world use cases for which people have been adopting them. In this work, we discuss how to adapt existing application-specific generation benchmarks to PLMs and provide an in-depth, empirical study of the limitations and capabilities of PLMs in natural language generation tasks along dimensions such as scale, architecture, input and output language. Our results show that PLMs differ in their applicability to different data regimes and their generalization to multiple languages. They further inform practitioners as to which PLMs to use for a given generation task setup. We share best practices to be taken into consideration when benchmarking generation capabilities during the development of upcoming PLMs.",
}
| Pre-trained large language models (PLMs) underly most new developments in natural language processing. They have shifted the field from application-specific model pipelines to a single model that is adapted to a wide range of tasks. Autoregressive PLMs like GPT-3 or PaLM and associated techniques like fewshot learning, have additionally shifted the output modality to generation instead of classification or regression. Despite their ubiquitous use, the generation quality of language models is rarely evaluated when these models are introduced. Additionally, it is unclear how existing generation tasks{--}while they can be used to compare systems at a high level{--}relate to the real world use cases for which people have been adopting them. In this work, we discuss how to adapt existing application-specific generation benchmarks to PLMs and provide an in-depth, empirical study of the limitations and capabilities of PLMs in natural language generation tasks along dimensions such as scale, architecture, input and output language. Our results show that PLMs differ in their applicability to different data regimes and their generalization to multiple languages. They further inform practitioners as to which PLMs to use for a given generation task setup. We share best practices to be taken into consideration when benchmarking generation capabilities during the development of upcoming PLMs. | [
"Maynez, Joshua",
"Agrawal, Priyanka",
"Gehrmann, Sebastian"
] | Benchmarking Large Language Model Capabilities for Conditional Generation | acl-long.511 | Poster | 2306.16793 | [
""
] | https://huggingface.co/papers/2306.16793 | 1 | 7 | 0 | 3 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.512.bib | https://aclanthology.org/2023.acl-long.512/ | @inproceedings{wu-etal-2023-lilgym,
title = "lil{G}ym: Natural Language Visual Reasoning with Reinforcement Learning",
author = "Wu, Anne and
Brantley, Kiante and
Kojima, Noriyuki and
Artzi, Yoav",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.512",
doi = "10.18653/v1/2023.acl-long.512",
pages = "9214--9234",
abstract = "We present lilGym, a new benchmark for language-conditioned reinforcement learning in visual environments. lilGym is based on 2,661 highly-compositional human-written natural language statements grounded in an interactive visual environment. We introduce a new approach for exact reward computation in every possible world state by annotating all statements with executable Python programs. Each statement is paired with multiple start states and reward functions to form thousands of distinct Markov Decision Processes of varying difficulty. We experiment with lilGym with different models and learning regimes. Our results and analysis show that while existing methods are able to achieve non-trivial performance, lilGym forms a challenging open problem. lilGym is available at \url{https://lil.nlp.cornell.edu/lilgym/}.",
}
| We present lilGym, a new benchmark for language-conditioned reinforcement learning in visual environments. lilGym is based on 2,661 highly-compositional human-written natural language statements grounded in an interactive visual environment. We introduce a new approach for exact reward computation in every possible world state by annotating all statements with executable Python programs. Each statement is paired with multiple start states and reward functions to form thousands of distinct Markov Decision Processes of varying difficulty. We experiment with lilGym with different models and learning regimes. Our results and analysis show that while existing methods are able to achieve non-trivial performance, lilGym forms a challenging open problem. lilGym is available at \url{https://lil.nlp.cornell.edu/lilgym/}. | [
"Wu, Anne",
"Brantley, Kiante",
"Kojima, Noriyuki",
"Artzi, Yoav"
] | lilGym: Natural Language Visual Reasoning with Reinforcement Learning | acl-long.512 | Poster | 2211.01994 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.513.bib | https://aclanthology.org/2023.acl-long.513/ | @inproceedings{tian-etal-2023-unsupervised,
title = "Unsupervised Melody-to-Lyrics Generation",
author = "Tian, Yufei and
Narayan-Chen, Anjali and
Oraby, Shereen and
Cervone, Alessandra and
Sigurdsson, Gunnar and
Tao, Chenyang and
Zhao, Wenbo and
Chen, Yiwen and
Chung, Tagyoung and
Huang, Jing and
Peng, Nanyun",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.513",
doi = "10.18653/v1/2023.acl-long.513",
pages = "9235--9254",
abstract = "Automatic melody-to-lyric generation is a task in which song lyrics are generated to go with a given melody. It is of significant practical interest and more challenging than unconstrained lyric generation as the music imposes additional constraints onto the lyrics. The training data is limited as most songs are copyrighted, resulting in models that underfit the complicated cross-modal relationship between melody and lyrics. In this work, we propose a method for generating high-quality lyrics without training on any aligned melody-lyric data. Specifically, we design a hierarchical lyric generation framework that first generates a song outline and second the complete lyrics. The framework enables disentanglement of training (based purely on text) from inference (melody-guided text generation) to circumvent the shortage of parallel data. We leverage the segmentation and rhythm alignment between melody and lyrics to compile the given melody into decoding constraints as guidance during inference. The two-step hierarchical design also enables content control via the lyric outline, a much-desired feature for democratizing collaborative song creation. Experimental results show that our model can generate high-quality lyrics that are more on-topic, singable, intelligible, and coherent than strong baselines, for example SongMASS, a SOTA model trained on a parallel dataset, with a 24{\%} relative overall quality improvement based on human ratings. Our code is available at \url{https://github.com/amazon-science/unsupervised-melody-to-lyrics-generation}.",
}
| Automatic melody-to-lyric generation is a task in which song lyrics are generated to go with a given melody. It is of significant practical interest and more challenging than unconstrained lyric generation as the music imposes additional constraints onto the lyrics. The training data is limited as most songs are copyrighted, resulting in models that underfit the complicated cross-modal relationship between melody and lyrics. In this work, we propose a method for generating high-quality lyrics without training on any aligned melody-lyric data. Specifically, we design a hierarchical lyric generation framework that first generates a song outline and second the complete lyrics. The framework enables disentanglement of training (based purely on text) from inference (melody-guided text generation) to circumvent the shortage of parallel data. We leverage the segmentation and rhythm alignment between melody and lyrics to compile the given melody into decoding constraints as guidance during inference. The two-step hierarchical design also enables content control via the lyric outline, a much-desired feature for democratizing collaborative song creation. Experimental results show that our model can generate high-quality lyrics that are more on-topic, singable, intelligible, and coherent than strong baselines, for example SongMASS, a SOTA model trained on a parallel dataset, with a 24{\%} relative overall quality improvement based on human ratings. Our code is available at \url{https://github.com/amazon-science/unsupervised-melody-to-lyrics-generation}. | [
"Tian, Yufei",
"Narayan-Chen, Anjali",
"Oraby, Shereen",
"Cervone, Aless",
"ra",
"Sigurdsson, Gunnar",
"Tao, Chenyang",
"Zhao, Wenbo",
"Chen, Yiwen",
"Chung, Tagyoung",
"Huang, Jing",
"Peng, Nanyun"
] | Unsupervised Melody-to-Lyrics Generation | acl-long.513 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.514.bib | https://aclanthology.org/2023.acl-long.514/ | @inproceedings{yuan-etal-2023-causality,
title = "Causality-aware Concept Extraction based on Knowledge-guided Prompting",
author = "Yuan, Siyu and
Yang, Deqing and
Liu, Jinxi and
Tian, Shuyu and
Liang, Jiaqing and
Xiao, Yanghua and
Xie, Rui",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.514",
doi = "10.18653/v1/2023.acl-long.514",
pages = "9255--9272",
abstract = "Concepts benefit natural language understanding but are far from complete in existing knowledge graphs (KGs). Recently, pre-trained language models (PLMs) have been widely used in text-based concept extraction (CE). However, PLMs tend to mine the co-occurrence associations from massive corpus as pre-trained knowledge rather than the real causal effect between tokens. As a result, the pre-trained knowledge confounds PLMs to extract biased concepts based on spurious co-occurrence correlations, inevitably resulting in low precision. In this paper, through the lens of a Structural Causal Model (SCM), we propose equipping the PLM-based extractor with a knowledge-guided prompt as an intervention to alleviate concept bias. The prompt adopts the topic of the given entity from the existing knowledge in KGs to mitigate the spurious co-occurrence correlations between entities and biased concepts. Our extensive experiments on representative multilingual KG datasets justify that our proposed prompt can effectively alleviate concept bias and improve the performance of PLM-based CE models.",
}
| Concepts benefit natural language understanding but are far from complete in existing knowledge graphs (KGs). Recently, pre-trained language models (PLMs) have been widely used in text-based concept extraction (CE). However, PLMs tend to mine the co-occurrence associations from massive corpus as pre-trained knowledge rather than the real causal effect between tokens. As a result, the pre-trained knowledge confounds PLMs to extract biased concepts based on spurious co-occurrence correlations, inevitably resulting in low precision. In this paper, through the lens of a Structural Causal Model (SCM), we propose equipping the PLM-based extractor with a knowledge-guided prompt as an intervention to alleviate concept bias. The prompt adopts the topic of the given entity from the existing knowledge in KGs to mitigate the spurious co-occurrence correlations between entities and biased concepts. Our extensive experiments on representative multilingual KG datasets justify that our proposed prompt can effectively alleviate concept bias and improve the performance of PLM-based CE models. | [
"Yuan, Siyu",
"Yang, Deqing",
"Liu, Jinxi",
"Tian, Shuyu",
"Liang, Jiaqing",
"Xiao, Yanghua",
"Xie, Rui"
] | Causality-aware Concept Extraction based on Knowledge-guided Prompting | acl-long.514 | Poster | 2305.01876 | [
"https://github.com/siyuyuan/kpce"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.515.bib | https://aclanthology.org/2023.acl-long.515/ | @inproceedings{zhang-etal-2023-span,
title = "Span-level Aspect-based Sentiment Analysis via Table Filling",
author = "Zhang, Mao and
Zhu, Yongxin and
Liu, Zhen and
Bao, Zhimin and
Wu, Yunfei and
Sun, Xing and
Xu, Linli",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.515",
doi = "10.18653/v1/2023.acl-long.515",
pages = "9273--9284",
abstract = "In this paper, we propose a novel span-level model for Aspect-Based Sentiment Analysis (ABSA), which aims at identifying the sentiment polarity of the given aspect. In contrast to conventional ABSA models that focus on modeling the word-level dependencies between an aspect and its corresponding opinion expressions, in this paper, we propose Table Filling BERT (TF-BERT), which considers the consistency of multi-word opinion expressions at the span-level. Specially, we learn the span representations with a table filling method, by constructing an upper triangular table for each sentiment polarity, of which the elements represent the sentiment intensity of the specific sentiment polarity for all spans in the sentence. Two methods are then proposed, including table-decoding and table-aggregation, to filter out target spans or aggregate each table for sentiment polarity classification. In addition, we design a sentiment consistency regularizer to guarantee the sentiment consistency of each span for different sentiment polarities. Experimental results on three benchmarks demonstrate the effectiveness of our proposed model.",
}
| In this paper, we propose a novel span-level model for Aspect-Based Sentiment Analysis (ABSA), which aims at identifying the sentiment polarity of the given aspect. In contrast to conventional ABSA models that focus on modeling the word-level dependencies between an aspect and its corresponding opinion expressions, in this paper, we propose Table Filling BERT (TF-BERT), which considers the consistency of multi-word opinion expressions at the span-level. Specially, we learn the span representations with a table filling method, by constructing an upper triangular table for each sentiment polarity, of which the elements represent the sentiment intensity of the specific sentiment polarity for all spans in the sentence. Two methods are then proposed, including table-decoding and table-aggregation, to filter out target spans or aggregate each table for sentiment polarity classification. In addition, we design a sentiment consistency regularizer to guarantee the sentiment consistency of each span for different sentiment polarities. Experimental results on three benchmarks demonstrate the effectiveness of our proposed model. | [
"Zhang, Mao",
"Zhu, Yongxin",
"Liu, Zhen",
"Bao, Zhimin",
"Wu, Yunfei",
"Sun, Xing",
"Xu, Linli"
] | Span-level Aspect-based Sentiment Analysis via Table Filling | acl-long.515 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.516.bib | https://aclanthology.org/2023.acl-long.516/ | @inproceedings{qian-etal-2023-limitations,
title = "Limitations of Language Models in Arithmetic and Symbolic Induction",
author = "Qian, Jing and
Wang, Hong and
Li, Zekun and
Li, Shiyang and
Yan, Xifeng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.516",
doi = "10.18653/v1/2023.acl-long.516",
pages = "9285--9298",
abstract = "Recent work has shown that large pretrained Language Models (LMs) can not only perform remarkably well on a range of Natural Language Processing (NLP) tasks but also start improving on reasoning tasks such as arithmetic induction, symbolic manipulation, and commonsense reasoning with increasing size of models. However, it is still unclear what the underlying capabilities of these LMs are. Surprisingly, we find that these models have limitations on certain basic symbolic manipulation tasks such as copy, reverse, and addition. When the total number of symbols or repeating symbols increases, the model performance drops quickly. We investigate the potential causes behind this phenomenon and examine a set of possible methods, including explicit positional markers, fine-grained computation steps, and LMs with callable programs. Experimental results show that none of these techniques can solve the simplest addition induction problem completely. In the end, we introduce LMs with tutor, which demonstrates every single step of teaching. LMs with tutor is able to deliver 100{\%} accuracy in situations of OOD and repeating symbols, shedding new insights on the boundary of large LMs in induction.",
}
| Recent work has shown that large pretrained Language Models (LMs) can not only perform remarkably well on a range of Natural Language Processing (NLP) tasks but also start improving on reasoning tasks such as arithmetic induction, symbolic manipulation, and commonsense reasoning with increasing size of models. However, it is still unclear what the underlying capabilities of these LMs are. Surprisingly, we find that these models have limitations on certain basic symbolic manipulation tasks such as copy, reverse, and addition. When the total number of symbols or repeating symbols increases, the model performance drops quickly. We investigate the potential causes behind this phenomenon and examine a set of possible methods, including explicit positional markers, fine-grained computation steps, and LMs with callable programs. Experimental results show that none of these techniques can solve the simplest addition induction problem completely. In the end, we introduce LMs with tutor, which demonstrates every single step of teaching. LMs with tutor is able to deliver 100{\%} accuracy in situations of OOD and repeating symbols, shedding new insights on the boundary of large LMs in induction. | [
"Qian, Jing",
"Wang, Hong",
"Li, Zekun",
"Li, Shiyang",
"Yan, Xifeng"
] | Limitations of Language Models in Arithmetic and Symbolic Induction | acl-long.516 | Poster | 2208.05051 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.517.bib | https://aclanthology.org/2023.acl-long.517/ | @inproceedings{singhal-etal-2023-eel,
title = "{EEL}: Efficiently Encoding Lattices for Reranking",
author = "Singhal, Prasann and
Xu, Jiacheng and
Ye, Xi and
Durrett, Greg",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.517",
doi = "10.18653/v1/2023.acl-long.517",
pages = "9299--9316",
abstract = "Standard decoding approaches for conditional text generation tasks typically search for an output hypothesis with high model probability, but this may not yield the best hypothesis according to human judgments of quality. Reranking to optimize for {``}downstream{''} metrics can more closely optimize for quality, but many metrics of interest are computed with pre-trained language models, which are slow to apply to large numbers of hypotheses. We explore an approach for reranking hypotheses by using Transformers to efficiently encode lattices of generated outputs, a method we call EEL. With a single Transformer pass over the entire lattice, we can approximately compute a contextualized representation of each token as if it were only part of a single hypothesis in isolation. We combine this approach with a new class of token-factored rerankers (TFRs) that allow for efficient extraction of high reranker-scoring hypotheses from the lattice. Empirically, our approach incurs minimal degradation error compared to the exponentially slower approach of encoding each hypothesis individually. When applying EEL with TFRs across three text generation tasks, our results show both substantial speedup compared to naive reranking and often better performance on downstream metrics than comparable approaches.",
}
| Standard decoding approaches for conditional text generation tasks typically search for an output hypothesis with high model probability, but this may not yield the best hypothesis according to human judgments of quality. Reranking to optimize for {``}downstream{''} metrics can more closely optimize for quality, but many metrics of interest are computed with pre-trained language models, which are slow to apply to large numbers of hypotheses. We explore an approach for reranking hypotheses by using Transformers to efficiently encode lattices of generated outputs, a method we call EEL. With a single Transformer pass over the entire lattice, we can approximately compute a contextualized representation of each token as if it were only part of a single hypothesis in isolation. We combine this approach with a new class of token-factored rerankers (TFRs) that allow for efficient extraction of high reranker-scoring hypotheses from the lattice. Empirically, our approach incurs minimal degradation error compared to the exponentially slower approach of encoding each hypothesis individually. When applying EEL with TFRs across three text generation tasks, our results show both substantial speedup compared to naive reranking and often better performance on downstream metrics than comparable approaches. | [
"Singhal, Prasann",
"Xu, Jiacheng",
"Ye, Xi",
"Durrett, Greg"
] | EEL: Efficiently Encoding Lattices for Reranking | acl-long.517 | Poster | 2306.00947 | [
"https://github.com/prasanns/eel-reranking"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.518.bib | https://aclanthology.org/2023.acl-long.518/ | @inproceedings{ye-etal-2023-clapspeech,
title = "{CLAPS}peech: Learning Prosody from Text Context with Contrastive Language-Audio Pre-Training",
author = "Ye, Zhenhui and
Huang, Rongjie and
Ren, Yi and
Jiang, Ziyue and
Liu, Jinglin and
He, Jinzheng and
Yin, Xiang and
Zhao, Zhou",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.518",
doi = "10.18653/v1/2023.acl-long.518",
pages = "9317--9331",
abstract = "Improving text representation has attracted much attention to achieve expressive text-to-speech (TTS). However, existing works only implicitly learn the prosody with masked token reconstruction tasks, which leads to low training efficiency and difficulty in prosody modeling. We propose CLAPSpeech, a cross-modal contrastive pre-training framework that learns from the prosody variance of the same text token under different contexts. Specifically, 1) with the design of a text encoder and a prosody encoder, we encourage the model to connect the text context with its corresponding prosody pattern in the joint multi-modal space; 2) we introduce a multi-scale pre-training pipeline to capture prosody patterns in multiple levels. 3) we show how to incorporate CLAPSpeech into existing TTS models for better prosody. Experiments on three datasets not only show that CLAPSpeech could improve the prosody prediction for existing TTS methods, but also demonstrate its generalization ability to adapt to multiple languages and multi-speaker text-to-speech. We also deeply analyze the principle behind the performance of CLAPSpeech. Ablation studies demonstrate the necessity of each component in CLAPSpeech. Source code and audio samples are available at \url{https://clapspeech.github.io}.",
}
| Improving text representation has attracted much attention to achieve expressive text-to-speech (TTS). However, existing works only implicitly learn the prosody with masked token reconstruction tasks, which leads to low training efficiency and difficulty in prosody modeling. We propose CLAPSpeech, a cross-modal contrastive pre-training framework that learns from the prosody variance of the same text token under different contexts. Specifically, 1) with the design of a text encoder and a prosody encoder, we encourage the model to connect the text context with its corresponding prosody pattern in the joint multi-modal space; 2) we introduce a multi-scale pre-training pipeline to capture prosody patterns in multiple levels. 3) we show how to incorporate CLAPSpeech into existing TTS models for better prosody. Experiments on three datasets not only show that CLAPSpeech could improve the prosody prediction for existing TTS methods, but also demonstrate its generalization ability to adapt to multiple languages and multi-speaker text-to-speech. We also deeply analyze the principle behind the performance of CLAPSpeech. Ablation studies demonstrate the necessity of each component in CLAPSpeech. Source code and audio samples are available at \url{https://clapspeech.github.io}. | [
"Ye, Zhenhui",
"Huang, Rongjie",
"Ren, Yi",
"Jiang, Ziyue",
"Liu, Jinglin",
"He, Jinzheng",
"Yin, Xiang",
"Zhao, Zhou"
] | CLAPSpeech: Learning Prosody from Text Context with Contrastive Language-Audio Pre-Training | acl-long.518 | Poster | 2305.10763 | [
""
] | https://huggingface.co/papers/2305.10763 | 3 | 3 | 4 | 8 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.519.bib | https://aclanthology.org/2023.acl-long.519/ | @inproceedings{chen-etal-2023-revisiting,
title = "Revisiting Cross-Lingual Summarization: A Corpus-based Study and A New Benchmark with Improved Annotation",
author = "Chen, Yulong and
Zhang, Huajian and
Zhou, Yijie and
Bai, Xuefeng and
Wang, Yueguan and
Zhong, Ming and
Yan, Jianhao and
Li, Yafu and
Li, Judy and
Zhu, Xianchao and
Zhang, Yue",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.519",
doi = "10.18653/v1/2023.acl-long.519",
pages = "9332--9351",
abstract = "Most existing cross-lingual summarization (CLS) work constructs CLS corpora by simply and directly translating pre-annotated summaries from one language to another, which can contain errors from both summarization and translation processes. To address this issue, we propose ConvSumX, a cross-lingual conversation summarization benchmark, through a new annotation schema that explicitly considers source input context. ConvSumX consists of 2 sub-tasks under different real-world scenarios, with each covering 3 language directions. We conduct thorough analysis on ConvSumX and 3 widely-used manually annotated CLS corpora and empirically find that ConvSumX is more faithful towards input text. Additionally, based on the same intuition, we propose a 2-Step method, which takes both conversation and summary as input to simulate human annotation process. Experimental results show that 2-Step method surpasses strong baselines on ConvSumX under both automatic and human evaluation. Analysis shows that both source input text and summary are crucial for modeling cross-lingual summaries.",
}
| Most existing cross-lingual summarization (CLS) work constructs CLS corpora by simply and directly translating pre-annotated summaries from one language to another, which can contain errors from both summarization and translation processes. To address this issue, we propose ConvSumX, a cross-lingual conversation summarization benchmark, through a new annotation schema that explicitly considers source input context. ConvSumX consists of 2 sub-tasks under different real-world scenarios, with each covering 3 language directions. We conduct thorough analysis on ConvSumX and 3 widely-used manually annotated CLS corpora and empirically find that ConvSumX is more faithful towards input text. Additionally, based on the same intuition, we propose a 2-Step method, which takes both conversation and summary as input to simulate human annotation process. Experimental results show that 2-Step method surpasses strong baselines on ConvSumX under both automatic and human evaluation. Analysis shows that both source input text and summary are crucial for modeling cross-lingual summaries. | [
"Chen, Yulong",
"Zhang, Huajian",
"Zhou, Yijie",
"Bai, Xuefeng",
"Wang, Yueguan",
"Zhong, Ming",
"Yan, Jianhao",
"Li, Yafu",
"Li, Judy",
"Zhu, Xianchao",
"Zhang, Yue"
] | Revisiting Cross-Lingual Summarization: A Corpus-based Study and A New Benchmark with Improved Annotation | acl-long.519 | Poster | 2307.04018 | [
"https://github.com/cylnlp/convsumx"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.520.bib | https://aclanthology.org/2023.acl-long.520/ | @inproceedings{tang-etal-2023-learning,
title = "Learning Dynamic Contextualised Word Embeddings via Template-based Temporal Adaptation",
author = "Tang, Xiaohang and
Zhou, Yi and
Bollegala, Danushka",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.520",
doi = "10.18653/v1/2023.acl-long.520",
pages = "9352--9369",
abstract = "Dynamic contextualised word embeddings (DCWEs) represent the temporal semantic variations of words. We propose a method for learning DCWEs by time-adapting a pretrained Masked Language Model (MLM) using time-sensitive templates. Given two snapshots $C_1$ and $C_2$ of a corpus taken respectively at two distinct timestamps $T_1$ and $T_2$, we first propose an unsupervised method to select (a) \textit{pivot} terms related to both $C_1$ and $C_2$, and (b) \textit{anchor} terms that are associated with a specific pivot term in each individual snapshot.We then generate prompts by filling manually compiled templates using the extracted pivot and anchor terms.Moreover, we propose an automatic method to learn time-sensitive templates from $C_1$ and $C_2$, without requiring any human supervision.Next, we use the generated prompts to adapt a pretrained MLM to $T_2$ by fine-tuning using those prompts.Multiple experiments show that our proposed method significantly reduces the perplexity of test sentences in $C_2$, outperforming the current state-of-the-art.",
}
| Dynamic contextualised word embeddings (DCWEs) represent the temporal semantic variations of words. We propose a method for learning DCWEs by time-adapting a pretrained Masked Language Model (MLM) using time-sensitive templates. Given two snapshots $C_1$ and $C_2$ of a corpus taken respectively at two distinct timestamps $T_1$ and $T_2$, we first propose an unsupervised method to select (a) \textit{pivot} terms related to both $C_1$ and $C_2$, and (b) \textit{anchor} terms that are associated with a specific pivot term in each individual snapshot.We then generate prompts by filling manually compiled templates using the extracted pivot and anchor terms.Moreover, we propose an automatic method to learn time-sensitive templates from $C_1$ and $C_2$, without requiring any human supervision.Next, we use the generated prompts to adapt a pretrained MLM to $T_2$ by fine-tuning using those prompts.Multiple experiments show that our proposed method significantly reduces the perplexity of test sentences in $C_2$, outperforming the current state-of-the-art. | [
"Tang, Xiaohang",
"Zhou, Yi",
"Bollegala, Danushka"
] | Learning Dynamic Contextualised Word Embeddings via Template-based Temporal Adaptation | acl-long.520 | Oral | 2208.10734 | [
"https://github.com/livnlp/timeadapted-dcwe"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.521.bib | https://aclanthology.org/2023.acl-long.521/ | @inproceedings{yedetore-etal-2023-poor,
title = "How poor is the stimulus? Evaluating hierarchical generalization in neural networks trained on child-directed speech",
author = "Yedetore, Aditya and
Linzen, Tal and
Frank, Robert and
McCoy, R. Thomas",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.521",
doi = "10.18653/v1/2023.acl-long.521",
pages = "9370--9393",
abstract = "When acquiring syntax, children consistently choose hierarchical rules over competing non-hierarchical possibilities. Is this preference due to a learning bias for hierarchical structure, or due to more general biases that interact with hierarchical cues in children{'}s linguistic input? We explore these possibilities by training LSTMs and Transformers - two types of neural networks without a hierarchical bias - on data similar in quantity and content to children{'}s linguistic input: text from the CHILDES corpus. We then evaluate what these models have learned about English yes/no questions, a phenomenon for which hierarchical structure is crucial. We find that, though they perform well at capturing the surface statistics of child-directed speech (as measured by perplexity), both model types generalize in a way more consistent with an incorrect linear rule than the correct hierarchical rule. These results suggest that human-like generalization from text alone requires stronger biases than the general sequence-processing biases of standard neural network architectures.",
}
| When acquiring syntax, children consistently choose hierarchical rules over competing non-hierarchical possibilities. Is this preference due to a learning bias for hierarchical structure, or due to more general biases that interact with hierarchical cues in children{'}s linguistic input? We explore these possibilities by training LSTMs and Transformers - two types of neural networks without a hierarchical bias - on data similar in quantity and content to children{'}s linguistic input: text from the CHILDES corpus. We then evaluate what these models have learned about English yes/no questions, a phenomenon for which hierarchical structure is crucial. We find that, though they perform well at capturing the surface statistics of child-directed speech (as measured by perplexity), both model types generalize in a way more consistent with an incorrect linear rule than the correct hierarchical rule. These results suggest that human-like generalization from text alone requires stronger biases than the general sequence-processing biases of standard neural network architectures. | [
"Yedetore, Aditya",
"Linzen, Tal",
"Frank, Robert",
"McCoy, R. Thomas"
] | How poor is the stimulus? Evaluating hierarchical generalization in neural networks trained on child-directed speech | acl-long.521 | Poster | 2301.11462 | [
"https://github.com/adityayedetore/lm-povstim-with-childes"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.522.bib | https://aclanthology.org/2023.acl-long.522/ | @inproceedings{yang-etal-2023-ganlm,
title = "{G}an{LM}: Encoder-Decoder Pre-training with an Auxiliary Discriminator",
author = "Yang, Jian and
Ma, Shuming and
Dong, Li and
Huang, Shaohan and
Huang, Haoyang and
Yin, Yuwei and
Zhang, Dongdong and
Yang, Liqun and
Wei, Furu and
Li, Zhoujun",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.522",
doi = "10.18653/v1/2023.acl-long.522",
pages = "9394--9412",
abstract = "Pre-trained models have achieved remarkable success in natural language processing (NLP). However, existing pre-training methods underutilize the benefits of language understanding for generation. Inspired by the idea of Generative Adversarial Networks (GANs), we propose a GAN-style model for encoder-decoder pre-training by introducing an auxiliary discriminator, unifying the ability of language understanding and generation in a single model. Our model, named as GanLM, is trained with two pre-training objectives: replaced token detection and replaced token denoising. Specifically, given masked source sentences, the generator outputs the target distribution and the discriminator predicts whether the target sampled tokens from distribution are incorrect. The target sentence is replaced with misclassified tokens to construct noisy previous context, which is used to generate the gold sentence. In general, both tasks improve the ability of language understanding and generation by selectively using the denoising data. Extensive experiments in language generation benchmarks show that GanLM with the powerful language understanding capability outperforms various strong pre-trained language models (PLMs) and achieves state-of-the-art performance.",
}
| Pre-trained models have achieved remarkable success in natural language processing (NLP). However, existing pre-training methods underutilize the benefits of language understanding for generation. Inspired by the idea of Generative Adversarial Networks (GANs), we propose a GAN-style model for encoder-decoder pre-training by introducing an auxiliary discriminator, unifying the ability of language understanding and generation in a single model. Our model, named as GanLM, is trained with two pre-training objectives: replaced token detection and replaced token denoising. Specifically, given masked source sentences, the generator outputs the target distribution and the discriminator predicts whether the target sampled tokens from distribution are incorrect. The target sentence is replaced with misclassified tokens to construct noisy previous context, which is used to generate the gold sentence. In general, both tasks improve the ability of language understanding and generation by selectively using the denoising data. Extensive experiments in language generation benchmarks show that GanLM with the powerful language understanding capability outperforms various strong pre-trained language models (PLMs) and achieves state-of-the-art performance. | [
"Yang, Jian",
"Ma, Shuming",
"Dong, Li",
"Huang, Shaohan",
"Huang, Haoyang",
"Yin, Yuwei",
"Zhang, Dongdong",
"Yang, Liqun",
"Wei, Furu",
"Li, Zhoujun"
] | GanLM: Encoder-Decoder Pre-training with an Auxiliary Discriminator | acl-long.522 | Poster | 2212.10218 | [
"https://github.com/csjianyang/ganlm"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.523.bib | https://aclanthology.org/2023.acl-long.523/ | @inproceedings{ravfogel-etal-2023-linear,
title = "Log-linear Guardedness and its Implications",
author = "Ravfogel, Shauli and
Goldberg, Yoav and
Cotterell, Ryan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.523",
doi = "10.18653/v1/2023.acl-long.523",
pages = "9413--9431",
abstract = "Methods for erasing human-interpretable concepts from neural representations that assume linearity have been found to be tractable and useful. However, the impact of this removal on the behavior of downstream classifiers trained on the modified representations is not fully understood. In this work, we formally define the notion of linear guardedness as the inability of an adversary to predict the concept directly from the representation, and study its implications. We show that, in the binary case, under certain assumptions, a downstream log-linear model cannot recover the erased concept. However, we constructively demonstrate that a multiclass log-linear model \textit{can} be constructed that indirectly recovers the concept in some cases, pointing to the inherent limitations of linear guardedness as a downstream bias mitigation technique.These findings shed light on the theoretical limitations of linear erasure methods and highlight the need for further research on the connections between intrinsic and extrinsic bias in neural models.",
}
| Methods for erasing human-interpretable concepts from neural representations that assume linearity have been found to be tractable and useful. However, the impact of this removal on the behavior of downstream classifiers trained on the modified representations is not fully understood. In this work, we formally define the notion of linear guardedness as the inability of an adversary to predict the concept directly from the representation, and study its implications. We show that, in the binary case, under certain assumptions, a downstream log-linear model cannot recover the erased concept. However, we constructively demonstrate that a multiclass log-linear model \textit{can} be constructed that indirectly recovers the concept in some cases, pointing to the inherent limitations of linear guardedness as a downstream bias mitigation technique.These findings shed light on the theoretical limitations of linear erasure methods and highlight the need for further research on the connections between intrinsic and extrinsic bias in neural models. | [
"Ravfogel, Shauli",
"Goldberg, Yoav",
"Cotterell, Ryan"
] | Log-linear Guardedness and its Implications | acl-long.523 | Poster | 2210.10012 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.524.bib | https://aclanthology.org/2023.acl-long.524/ | @inproceedings{briakou-etal-2023-searching,
title = "Searching for Needles in a Haystack: On the Role of Incidental Bilingualism in {P}a{LM}{'}s Translation Capability",
author = "Briakou, Eleftheria and
Cherry, Colin and
Foster, George",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.524",
doi = "10.18653/v1/2023.acl-long.524",
pages = "9432--9452",
abstract = "Large, multilingual language models exhibit surprisingly good zero- or few-shot machine translation capabilities, despite having never seen the intentionally-included translation examples provided to typical neural translation systems. We investigate the role of incidental bilingualism{---}the unintentional consumption of bilingual signals, including translation examples{---}in explaining the translation capabilities of large language models, taking the Pathways Language Model (PaLM) as a case study. We introduce a mixed-method approach to measure and understand incidental bilingualism at scale. We show that PaLM is exposed to over 30 million translation pairs across at least 44 languages. Furthermore, the amount of incidental bilingual content is highly correlated with the amount of monolingual in-language content for non-English languages. We relate incidental bilingual content to zero-shot prompts and show that it can be used to mine new prompts to improve PaLM{'}s out-of-English zero-shot translation quality. Finally, in a series of small-scale ablations, we show that its presence has a substantial impact on translation capabilities, although this impact diminishes with model scale.",
}
| Large, multilingual language models exhibit surprisingly good zero- or few-shot machine translation capabilities, despite having never seen the intentionally-included translation examples provided to typical neural translation systems. We investigate the role of incidental bilingualism{---}the unintentional consumption of bilingual signals, including translation examples{---}in explaining the translation capabilities of large language models, taking the Pathways Language Model (PaLM) as a case study. We introduce a mixed-method approach to measure and understand incidental bilingualism at scale. We show that PaLM is exposed to over 30 million translation pairs across at least 44 languages. Furthermore, the amount of incidental bilingual content is highly correlated with the amount of monolingual in-language content for non-English languages. We relate incidental bilingual content to zero-shot prompts and show that it can be used to mine new prompts to improve PaLM{'}s out-of-English zero-shot translation quality. Finally, in a series of small-scale ablations, we show that its presence has a substantial impact on translation capabilities, although this impact diminishes with model scale. | [
"Briakou, Eleftheria",
"Cherry, Colin",
"Foster, George"
] | Searching for Needles in a Haystack: On the Role of Incidental Bilingualism in PaLM's Translation Capability | acl-long.524 | Oral | 2305.10266 | [
""
] | https://huggingface.co/papers/2305.10266 | 2 | 1 | 0 | 3 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.525.bib | https://aclanthology.org/2023.acl-long.525/ | @inproceedings{zhao-etal-2023-open,
title = "Open Set Relation Extraction via Unknown-Aware Training",
author = "Zhao, Jun and
Zhao, Xin and
Zhan, WenYu and
Zhang, Qi and
Gui, Tao and
Wei, Zhongyu and
Chen, Yun Wen and
Gao, Xiang and
Huang, Xuanjing",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.525",
doi = "10.18653/v1/2023.acl-long.525",
pages = "9453--9467",
abstract = "The existing supervised relation extraction methods have achieved impressive performance in a closed-set setting, in which the relations remain the same during both training and testing. In a more realistic open-set setting, unknown relations may appear in the test set. Due to the lack of supervision signals from unknown relations, a well-performing closed-set relation extractor can still confidently misclassify them into known relations. In this paper, we propose an unknown-aware training method, regularizing the model by dynamically synthesizing negative instances that can provide the missing supervision signals. Inspired by text adversarial attack, We adaptively apply small but critical perturbations to original training data,synthesizing \textbf{difficult enough} negative instances that are mistaken by the model as known relations, thus facilitating a compact decision boundary. Experimental results show that our method achieves SOTA unknown relation detection without compromising the classification of known relations.",
}
| The existing supervised relation extraction methods have achieved impressive performance in a closed-set setting, in which the relations remain the same during both training and testing. In a more realistic open-set setting, unknown relations may appear in the test set. Due to the lack of supervision signals from unknown relations, a well-performing closed-set relation extractor can still confidently misclassify them into known relations. In this paper, we propose an unknown-aware training method, regularizing the model by dynamically synthesizing negative instances that can provide the missing supervision signals. Inspired by text adversarial attack, We adaptively apply small but critical perturbations to original training data,synthesizing \textbf{difficult enough} negative instances that are mistaken by the model as known relations, thus facilitating a compact decision boundary. Experimental results show that our method achieves SOTA unknown relation detection without compromising the classification of known relations. | [
"Zhao, Jun",
"Zhao, Xin",
"Zhan, WenYu",
"Zhang, Qi",
"Gui, Tao",
"Wei, Zhongyu",
"Chen, Yun Wen",
"Gao, Xiang",
"Huang, Xuanjing"
] | Open Set Relation Extraction via Unknown-Aware Training | acl-long.525 | Oral | 2306.04950 | [
"https://github.com/xinzhao0211/opensetre"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.526.bib | https://aclanthology.org/2023.acl-long.526/ | @inproceedings{tang-etal-2023-learning-imagine,
title = "Learning to Imagine: Visually-Augmented Natural Language Generation",
author = "Tang, Tianyi and
Chen, Yushuo and
Du, Yifan and
Li, Junyi and
Zhao, Wayne Xin and
Wen, Ji-Rong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.526",
doi = "10.18653/v1/2023.acl-long.526",
pages = "9468--9481",
abstract = "People often imagine relevant scenes to aid in the writing process. In this work, we aim to utilize visual information for composition in the same manner as humans. We propose a method, LIVE, that makes pre-trained language models (PLMs) Learn to Imagine for Visually-augmented natural language gEneration. First, we imagine the scene based on the text: we use a diffusion model to synthesize high-quality images conditioned on the input texts. Second, we use CLIP to determine whether the text can evoke the imagination in a posterior way. Finally, our imagination is dynamic, and we conduct synthesis for each sentence rather than generate only one image for an entire paragraph. Technically, we propose a novel plug-and-play fusion layer to obtain visually-augmented representations for each text. Our vision-text fusion layer is compatible with Transformer-based architecture. We have conducted extensive experiments on four generation tasks using BART and T5, and the automatic results and human evaluation demonstrate the effectiveness of our proposed method. We will release the code, model, and data at the link: \url{https://github.com/RUCAIBox/LIVE}.",
}
| People often imagine relevant scenes to aid in the writing process. In this work, we aim to utilize visual information for composition in the same manner as humans. We propose a method, LIVE, that makes pre-trained language models (PLMs) Learn to Imagine for Visually-augmented natural language gEneration. First, we imagine the scene based on the text: we use a diffusion model to synthesize high-quality images conditioned on the input texts. Second, we use CLIP to determine whether the text can evoke the imagination in a posterior way. Finally, our imagination is dynamic, and we conduct synthesis for each sentence rather than generate only one image for an entire paragraph. Technically, we propose a novel plug-and-play fusion layer to obtain visually-augmented representations for each text. Our vision-text fusion layer is compatible with Transformer-based architecture. We have conducted extensive experiments on four generation tasks using BART and T5, and the automatic results and human evaluation demonstrate the effectiveness of our proposed method. We will release the code, model, and data at the link: \url{https://github.com/RUCAIBox/LIVE}. | [
"Tang, Tianyi",
"Chen, Yushuo",
"Du, Yifan",
"Li, Junyi",
"Zhao, Wayne Xin",
"Wen, Ji-Rong"
] | Learning to Imagine: Visually-Augmented Natural Language Generation | acl-long.526 | Oral | 2305.16944 | [
"https://github.com/rucaibox/live"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.527.bib | https://aclanthology.org/2023.acl-long.527/ | @inproceedings{yu-etal-2023-generating,
title = "Generating Hashtags for Short-form Videos with Guided Signals",
author = "Yu, Tiezheng and
Yu, Hanchao and
Liang, Davis and
Mao, Yuning and
Nie, Shaoliang and
Huang, Po-Yao and
Khabsa, Madian and
Fung, Pascale and
Wang, Yi-Chia",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.527",
doi = "10.18653/v1/2023.acl-long.527",
pages = "9482--9495",
abstract = "Short-form video hashtag recommendation (SVHR) aims to recommend hashtags to content creators from videos and corresponding descriptions. Most prior studies regard SVHR as a classification or ranking problem and select hashtags from a set of limited candidates. However, in reality, users can create new hashtags, and trending hashtags change rapidly over time on social media. Both of these properties cannot be easily modeled with classification approaches. To bridge this gap, we formulate SVHR as a generation task that better represents how hashtags are created naturally. Additionally, we propose the Guided Generative Model (GGM) where we augment the input features by retrieving relevant hashtags from a large-scale hashtag pool as extra guidance signals. Experimental results on two short-form video datasets show that our generative models outperform strong classification baselines, and the guidance signals further boost the performance by 8.11 and 2.17 absolute ROUGE-1 scores on average, respectively. We also perform extensive analyses including human evaluation, demonstrating that our generative model can create meaningful and relevant novel hashtags while achieving state-of-the-art performance on known hashtags",
}
| Short-form video hashtag recommendation (SVHR) aims to recommend hashtags to content creators from videos and corresponding descriptions. Most prior studies regard SVHR as a classification or ranking problem and select hashtags from a set of limited candidates. However, in reality, users can create new hashtags, and trending hashtags change rapidly over time on social media. Both of these properties cannot be easily modeled with classification approaches. To bridge this gap, we formulate SVHR as a generation task that better represents how hashtags are created naturally. Additionally, we propose the Guided Generative Model (GGM) where we augment the input features by retrieving relevant hashtags from a large-scale hashtag pool as extra guidance signals. Experimental results on two short-form video datasets show that our generative models outperform strong classification baselines, and the guidance signals further boost the performance by 8.11 and 2.17 absolute ROUGE-1 scores on average, respectively. We also perform extensive analyses including human evaluation, demonstrating that our generative model can create meaningful and relevant novel hashtags while achieving state-of-the-art performance on known hashtags | [
"Yu, Tiezheng",
"Yu, Hanchao",
"Liang, Davis",
"Mao, Yuning",
"Nie, Shaoliang",
"Huang, Po-Yao",
"Khabsa, Madian",
"Fung, Pascale",
"Wang, Yi-Chia"
] | Generating Hashtags for Short-form Videos with Guided Signals | acl-long.527 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.528.bib | https://aclanthology.org/2023.acl-long.528/ | @inproceedings{bastan-etal-2023-neurostructural,
title = "{NEUROSTRUCTURAL} {DECODING}: Neural Text Generation with Structural Constraints",
author = "Bastan, Mohaddeseh and
Surdeanu, Mihai and
Balasubramanian, Niranjan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.528",
doi = "10.18653/v1/2023.acl-long.528",
pages = "9496--9510",
abstract = "Text generation often involves producing coherent and grammatically correct texts that also satisfy a given set of semantic constraints. While most approaches for conditional text generation have primarily focused on lexical constraints, they often struggle to effectively incorporate syntactic constraints, which provide a richer language for approximating semantic constraints. We address this gap by introducing NeuroStructural Decoding, a new decoding algorithm that incorporates syntactic constraints to further improve the quality of the generated text. We build NeuroStructural Decoding on the NeuroLogic Decoding (Lu etal. 2021) algorithm, which enables language generation models to produce fluent text while satisfying complex lexical constraints. Our algorithm is powerful and scalable. It tracks lexico-syntactic constraints (e.g., we need to observe dog as subject and ball as object)during decoding by parsing the partial generations at each step. To this end, we adapt a dependency parser to generate parses for incomplete sentences. Our approach is evaluated on three different language generation tasks, and the results show improved performance in both lexical and syntactic metrics compared to previous methods. The results suggest this is a promising solution for integrating fine-grained controllable generation into the conventional beam search decoding.",
}
| Text generation often involves producing coherent and grammatically correct texts that also satisfy a given set of semantic constraints. While most approaches for conditional text generation have primarily focused on lexical constraints, they often struggle to effectively incorporate syntactic constraints, which provide a richer language for approximating semantic constraints. We address this gap by introducing NeuroStructural Decoding, a new decoding algorithm that incorporates syntactic constraints to further improve the quality of the generated text. We build NeuroStructural Decoding on the NeuroLogic Decoding (Lu etal. 2021) algorithm, which enables language generation models to produce fluent text while satisfying complex lexical constraints. Our algorithm is powerful and scalable. It tracks lexico-syntactic constraints (e.g., we need to observe dog as subject and ball as object)during decoding by parsing the partial generations at each step. To this end, we adapt a dependency parser to generate parses for incomplete sentences. Our approach is evaluated on three different language generation tasks, and the results show improved performance in both lexical and syntactic metrics compared to previous methods. The results suggest this is a promising solution for integrating fine-grained controllable generation into the conventional beam search decoding. | [
"Bastan, Mohaddeseh",
"Surdeanu, Mihai",
"Balasubramanian, Niranjan"
] | NEUROSTRUCTURAL DECODING: Neural Text Generation with Structural Constraints | acl-long.528 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.529.bib | https://aclanthology.org/2023.acl-long.529/ | @inproceedings{li-etal-2023-best,
title = "The Best of Both Worlds: Combining Human and Machine Translations for Multilingual Semantic Parsing with Active Learning",
author = "Li, Zhuang and
Qu, Lizhen and
Cohen, Philip and
Tumuluri, Raj and
Haffari, Gholamreza",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.529",
doi = "10.18653/v1/2023.acl-long.529",
pages = "9511--9528",
abstract = "Multilingual semantic parsing aims to leverage the knowledge from the high-resource languages to improve low-resource semantic parsing, yet commonly suffers from the data imbalance problem. Prior works propose to utilize the translations by either humans or machines to alleviate such issues. However, human translations are expensive, while machine translations are cheap but prone to error and bias. In this work, we propose an active learning approach that exploits the strengths of both human and machine translations by iteratively adding small batches of human translations into the machine-translated training set. Besides, we propose novel aggregated acquisition criteria that help our active learning method select utterances to be manually translated. Our experiments demonstrate that an ideal utterance selection can significantly reduce the error and bias in the translated data, resulting in higher parser accuracies than the parsers merely trained on the machine-translated data.",
}
| Multilingual semantic parsing aims to leverage the knowledge from the high-resource languages to improve low-resource semantic parsing, yet commonly suffers from the data imbalance problem. Prior works propose to utilize the translations by either humans or machines to alleviate such issues. However, human translations are expensive, while machine translations are cheap but prone to error and bias. In this work, we propose an active learning approach that exploits the strengths of both human and machine translations by iteratively adding small batches of human translations into the machine-translated training set. Besides, we propose novel aggregated acquisition criteria that help our active learning method select utterances to be manually translated. Our experiments demonstrate that an ideal utterance selection can significantly reduce the error and bias in the translated data, resulting in higher parser accuracies than the parsers merely trained on the machine-translated data. | [
"Li, Zhuang",
"Qu, Lizhen",
"Cohen, Philip",
"Tumuluri, Raj",
"Haffari, Gholamreza"
] | The Best of Both Worlds: Combining Human and Machine Translations for Multilingual Semantic Parsing with Active Learning | acl-long.529 | Poster | 2305.12737 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.530.bib | https://aclanthology.org/2023.acl-long.530/ | @inproceedings{chen-etal-2023-ideology,
title = "Ideology Prediction from Scarce and Biased Supervision: Learn to Disregard the {``}What{''} and Focus on the {``}How{''}!",
author = "Chen, Chen and
Walker, Dylan and
Saligrama, Venkatesh",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.530",
doi = "10.18653/v1/2023.acl-long.530",
pages = "9529--9549",
abstract = "We propose a novel supervised learning approach for political ideology prediction (PIP) that is capable of predicting out-of-distribution inputs. This problem is motivated by the fact that manual data-labeling is expensive, while self-reported labels are often scarce and exhibit significant selection bias. We propose a novel statistical model that decomposes the document embeddings into a linear superposition of two vectors; a latent neutral \textit{context} vector independent of ideology, and a latent \textit{position} vector aligned with ideology. We train an end-to-end model that has intermediate contextual and positional vectors as outputs. At deployment time, our model predicts labels for input documents by exclusively leveraging the predicted positional vectors. On two benchmark datasets we show that our model is capable of outputting predictions even when trained with as little as 5{\%} biased data, and is significantly more accurate than the state-of-the-art. Through crowd-sourcing we validate the neutrality of contextual vectors, and show that context filtering results in ideological concentration, allowing for prediction on out-of-distribution examples.",
}
| We propose a novel supervised learning approach for political ideology prediction (PIP) that is capable of predicting out-of-distribution inputs. This problem is motivated by the fact that manual data-labeling is expensive, while self-reported labels are often scarce and exhibit significant selection bias. We propose a novel statistical model that decomposes the document embeddings into a linear superposition of two vectors; a latent neutral \textit{context} vector independent of ideology, and a latent \textit{position} vector aligned with ideology. We train an end-to-end model that has intermediate contextual and positional vectors as outputs. At deployment time, our model predicts labels for input documents by exclusively leveraging the predicted positional vectors. On two benchmark datasets we show that our model is capable of outputting predictions even when trained with as little as 5{\%} biased data, and is significantly more accurate than the state-of-the-art. Through crowd-sourcing we validate the neutrality of contextual vectors, and show that context filtering results in ideological concentration, allowing for prediction on out-of-distribution examples. | [
"Chen, Chen",
"Walker, Dylan",
"Saligrama, Venkatesh"
] | Ideology Prediction from Scarce and Biased Supervision: Learn to Disregard the “What” and Focus on the “How”! | acl-long.530 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.531.bib | https://aclanthology.org/2023.acl-long.531/ | @inproceedings{sosea-etal-2023-unsupervised,
title = "Unsupervised Extractive Summarization of Emotion Triggers",
author = "Sosea, Tiberiu and
Zhan, Hongli and
Li, Junyi Jessy and
Caragea, Cornelia",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.531",
doi = "10.18653/v1/2023.acl-long.531",
pages = "9550--9569",
abstract = "Understanding what leads to emotions during large-scale crises is important as it can provide groundings for expressed emotions and subsequently improve the understanding of ongoing disasters. Recent approaches trained supervised models to both detect emotions and explain emotion triggers (events and appraisals) via abstractive summarization. However, obtaining timely and qualitative abstractive summaries is expensive and extremely time-consuming, requiring highly-trained expert annotators. In time-sensitive, high-stake contexts, this can block necessary responses. We instead pursue unsupervised systems that extract triggers from text. First, we introduce CovidET-EXT, augmenting (Zhan et al., 2022){'}s abstractive dataset (in the context of the COVID-19 crisis) with extractive triggers. Second, we develop new unsupervised learning models that can jointly detect emotions and summarize their triggers. Our best approach, entitled Emotion-Aware Pagerank, incorporates emotion information from external sources combined with a language understanding module, and outperforms strong baselines. We release our data and code at \url{https://github.com/tsosea2/CovidET-EXT}.",
}
| Understanding what leads to emotions during large-scale crises is important as it can provide groundings for expressed emotions and subsequently improve the understanding of ongoing disasters. Recent approaches trained supervised models to both detect emotions and explain emotion triggers (events and appraisals) via abstractive summarization. However, obtaining timely and qualitative abstractive summaries is expensive and extremely time-consuming, requiring highly-trained expert annotators. In time-sensitive, high-stake contexts, this can block necessary responses. We instead pursue unsupervised systems that extract triggers from text. First, we introduce CovidET-EXT, augmenting (Zhan et al., 2022){'}s abstractive dataset (in the context of the COVID-19 crisis) with extractive triggers. Second, we develop new unsupervised learning models that can jointly detect emotions and summarize their triggers. Our best approach, entitled Emotion-Aware Pagerank, incorporates emotion information from external sources combined with a language understanding module, and outperforms strong baselines. We release our data and code at \url{https://github.com/tsosea2/CovidET-EXT}. | [
"Sosea, Tiberiu",
"Zhan, Hongli",
"Li, Junyi Jessy",
"Caragea, Cornelia"
] | Unsupervised Extractive Summarization of Emotion Triggers | acl-long.531 | Poster | 2306.01444 | [
"https://github.com/tsosea2/covidet-ext"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.532.bib | https://aclanthology.org/2023.acl-long.532/ | @inproceedings{liu-etal-2023-document,
title = "Document-Level Event Argument Extraction With a Chain Reasoning Paradigm",
author = "Liu, Jian and
Liang, Chen and
Xu, Jinan and
Liu, Haoyan and
Zhao, Zhe",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.532",
doi = "10.18653/v1/2023.acl-long.532",
pages = "9570--9583",
abstract = "Document-level event argument extraction aims to identify event arguments beyond sentence level, where a significant challenge is to model long-range dependencies. Focusing on this challenge, we present a new chain reasoning paradigm for the task, which can generate decomposable first-order logic rules for reasoning. This paradigm naturally captures long-range interdependence due to the chains{'} compositional nature, which also improves interpretability by explicitly modeling the reasoning process. We introduce T-norm fuzzy logic for optimization, which permits end-to-end learning and shows promise for integrating the expressiveness of logical reasoning with the generalization of neural networks. In experiments, we show that our approach outperforms previous methods by a significant margin on two standard benchmarks (over 6 points in F1).Moreover, it is data-efficient in low-resource scenarios and robust enough to defend against adversarial attacks.",
}
| Document-level event argument extraction aims to identify event arguments beyond sentence level, where a significant challenge is to model long-range dependencies. Focusing on this challenge, we present a new chain reasoning paradigm for the task, which can generate decomposable first-order logic rules for reasoning. This paradigm naturally captures long-range interdependence due to the chains{'} compositional nature, which also improves interpretability by explicitly modeling the reasoning process. We introduce T-norm fuzzy logic for optimization, which permits end-to-end learning and shows promise for integrating the expressiveness of logical reasoning with the generalization of neural networks. In experiments, we show that our approach outperforms previous methods by a significant margin on two standard benchmarks (over 6 points in F1).Moreover, it is data-efficient in low-resource scenarios and robust enough to defend against adversarial attacks. | [
"Liu, Jian",
"Liang, Chen",
"Xu, Jinan",
"Liu, Haoyan",
"Zhao, Zhe"
] | Document-Level Event Argument Extraction With a Chain Reasoning Paradigm | acl-long.532 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.533.bib | https://aclanthology.org/2023.acl-long.533/ | @inproceedings{li-etal-2023-pre,
title = "Pre-training Multi-party Dialogue Models with Latent Discourse Inference",
author = "Li, Yiyang and
Huang, Xinting and
Bi, Wei and
Zhao, Hai",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.533",
doi = "10.18653/v1/2023.acl-long.533",
pages = "9584--9599",
abstract = "Multi-party dialogues are more difficult for models to understand than one-to-one two-party dialogues, since they involve multiple interlocutors, resulting in interweaving reply-to relations and information flows. To step over these obstacles, an effective way is to pre-train a model that understands the discourse structure of multi-party dialogues, namely, to whom each utterance is replying. However, due to the lack of explicitly annotated discourse labels in multi-party dialogue corpora, previous works fail to scale up the pre-training process by putting aside the unlabeled multi-party conversational data for nothing. To fully utilize the unlabeled data, we propose to treat the discourse structures as latent variables, then jointly infer them and pre-train the discourse-aware model by unsupervised latent variable inference methods. Experiments on multiple downstream tasks show that our pre-trained model outperforms strong baselines by large margins and achieves state-of-the-art (SOTA) results, justifying the effectiveness of our method. The official implementation of this paper is available at \url{https://github.com/EricLee8/MPD_EMVI}.",
}
| Multi-party dialogues are more difficult for models to understand than one-to-one two-party dialogues, since they involve multiple interlocutors, resulting in interweaving reply-to relations and information flows. To step over these obstacles, an effective way is to pre-train a model that understands the discourse structure of multi-party dialogues, namely, to whom each utterance is replying. However, due to the lack of explicitly annotated discourse labels in multi-party dialogue corpora, previous works fail to scale up the pre-training process by putting aside the unlabeled multi-party conversational data for nothing. To fully utilize the unlabeled data, we propose to treat the discourse structures as latent variables, then jointly infer them and pre-train the discourse-aware model by unsupervised latent variable inference methods. Experiments on multiple downstream tasks show that our pre-trained model outperforms strong baselines by large margins and achieves state-of-the-art (SOTA) results, justifying the effectiveness of our method. The official implementation of this paper is available at \url{https://github.com/EricLee8/MPD_EMVI}. | [
"Li, Yiyang",
"Huang, Xinting",
"Bi, Wei",
"Zhao, Hai"
] | Pre-training Multi-party Dialogue Models with Latent Discourse Inference | acl-long.533 | Poster | 2305.15175 | [
"https://github.com/ericlee8/mpd_emvi"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.534.bib | https://aclanthology.org/2023.acl-long.534/ | @inproceedings{xilong-etal-2023-interpreting,
title = "Interpreting Positional Information in Perspective of Word Order",
author = "Xilong, Zhang and
Ruochen, Liu and
Jin, Liu and
Xuefeng, Liang",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.534",
doi = "10.18653/v1/2023.acl-long.534",
pages = "9600--9613",
abstract = "The attention mechanism is a powerful and effective method utilized in natural language processing. However, it has been observed that this method is insensitive to positional information. Although several studies have attempted to improve positional encoding and investigate the influence of word order perturbation, it remains unclear how positional encoding impacts NLP models from the perspective of word order. In this paper, we aim to shed light on this problem by analyzing the working mechanism of the attention module and investigating the root cause of its inability to encode positional information. Our hypothesis is that the insensitivity can be attributed to the weight sum operation utilized in the attention module. To verify this hypothesis, we propose a novel weight concatenation operation and evaluate its efficacy in neural machine translation tasks. Our enhanced experimental results not only reveal that the proposed operation can effectively encode positional information but also confirm our hypothesis.",
}
| The attention mechanism is a powerful and effective method utilized in natural language processing. However, it has been observed that this method is insensitive to positional information. Although several studies have attempted to improve positional encoding and investigate the influence of word order perturbation, it remains unclear how positional encoding impacts NLP models from the perspective of word order. In this paper, we aim to shed light on this problem by analyzing the working mechanism of the attention module and investigating the root cause of its inability to encode positional information. Our hypothesis is that the insensitivity can be attributed to the weight sum operation utilized in the attention module. To verify this hypothesis, we propose a novel weight concatenation operation and evaluate its efficacy in neural machine translation tasks. Our enhanced experimental results not only reveal that the proposed operation can effectively encode positional information but also confirm our hypothesis. | [
"Xilong, Zhang",
"Ruochen, Liu",
"Jin, Liu",
"Xuefeng, Liang"
] | Interpreting Positional Information in Perspective of Word Order | acl-long.534 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.535.bib | https://aclanthology.org/2023.acl-long.535/ | @inproceedings{bhagavatula-etal-2023-i2d2,
title = "{I}2{D}2: Inductive Knowledge Distillation with {N}euro{L}ogic and Self-Imitation",
author = "Bhagavatula, Chandra and
Hwang, Jena D. and
Downey, Doug and
Le Bras, Ronan and
Lu, Ximing and
Qin, Lianhui and
Sakaguchi, Keisuke and
Swayamdipta, Swabha and
West, Peter and
Choi, Yejin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.535",
doi = "10.18653/v1/2023.acl-long.535",
pages = "9614--9630",
abstract = "Commonsense capabilities of pre-trained language models dramatically improve with scale, leading many to believe that scale is the only winning recipe. But is it? Here, we investigate an alternative that a priori seems impossible: can smaller language models (e.g., GPT-2) win over models that are orders of magnitude larger and better (e.g., GPT-3), if powered with novel commonsense distillation algorithms?The key intellectual challenge is to design a learning algorithm that achieve a competitive level of commonsense acquisition, without relying on the benefits of scale. In particular, we study generative models of commonsense knowledge, focusing on the task of generating generics, statements of commonsense facts about everyday concepts, e.g., birds can fly. We introduce I2D2, a novel commonsense distillation framework that loosely follows the Symbolic Knowledge Distillation of West et al. but breaks the dependence on the extreme-scale teacher model with two innovations: (1) the novel adaptation of NeuroLogic Decoding to enhance the generation quality of the weak, off-the-shelf language models, and (2) self-imitation learning to iteratively learn from the model{'}s own enhanced commonsense acquisition capabilities. Empirical results suggest that scale is not the only way, as novel algorithms can be a promising alternative. Moreover, our study leads to a new corpus of generics, Gen-A-tomic, that is the largest and highest quality available to date.",
}
| Commonsense capabilities of pre-trained language models dramatically improve with scale, leading many to believe that scale is the only winning recipe. But is it? Here, we investigate an alternative that a priori seems impossible: can smaller language models (e.g., GPT-2) win over models that are orders of magnitude larger and better (e.g., GPT-3), if powered with novel commonsense distillation algorithms?The key intellectual challenge is to design a learning algorithm that achieve a competitive level of commonsense acquisition, without relying on the benefits of scale. In particular, we study generative models of commonsense knowledge, focusing on the task of generating generics, statements of commonsense facts about everyday concepts, e.g., birds can fly. We introduce I2D2, a novel commonsense distillation framework that loosely follows the Symbolic Knowledge Distillation of West et al. but breaks the dependence on the extreme-scale teacher model with two innovations: (1) the novel adaptation of NeuroLogic Decoding to enhance the generation quality of the weak, off-the-shelf language models, and (2) self-imitation learning to iteratively learn from the model{'}s own enhanced commonsense acquisition capabilities. Empirical results suggest that scale is not the only way, as novel algorithms can be a promising alternative. Moreover, our study leads to a new corpus of generics, Gen-A-tomic, that is the largest and highest quality available to date. | [
"Bhagavatula, Ch",
"ra",
"Hwang, Jena D.",
"Downey, Doug",
"Le Bras, Ronan",
"Lu, Ximing",
"Qin, Lianhui",
"Sakaguchi, Keisuke",
"Swayamdipta, Swabha",
"West, Peter",
"Choi, Yejin"
] | I2D2: Inductive Knowledge Distillation with NeuroLogic and Self-Imitation | acl-long.535 | Poster | 2212.09246 | [
""
] | https://huggingface.co/papers/2212.09246 | 0 | 0 | 0 | 10 | 1 | [] | [
"tasksource/I2D2"
] | [] |
https://aclanthology.org/2023.acl-long.536.bib | https://aclanthology.org/2023.acl-long.536/ | @inproceedings{huang-etal-2023-classification,
title = "More than Classification: A Unified Framework for Event Temporal Relation Extraction",
author = "Huang, Quzhe and
Hu, Yutong and
Zhu, Shengqi and
Feng, Yansong and
Liu, Chang and
Zhao, Dongyan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.536",
doi = "10.18653/v1/2023.acl-long.536",
pages = "9631--9646",
abstract = "Event temporal relation extraction (ETRE) is usually formulated as a multi-label classification task, where each type of relation is simply treated as a one-hot label. This formulation ignores the meaning of relations and wipes out their intrinsic dependency. After examining the relation definitions in various ETRE tasks, we observe that all relations can be interpreted using the start and end time points of events. For example, relation \textit{Includes} could be interpreted as event 1 starting no later than event 2 and ending no earlier than event 2. In this paper, we propose a unified event temporal relation extraction framework, which transforms temporal relations into logical expressions of time points and completes the ETRE by predicting the relations between certain time point pairs. Experiments on TB-Dense and MATRES show significant improvements over a strong baseline and outperform the state-of-the-art model by 0.3{\%} on both datasets. By representing all relations in a unified framework, we can leverage the relations with sufficient data to assist the learning of other relations, thus achieving stable improvement in low-data scenarios. When the relation definitions are changed, our method can quickly adapt to the new ones by simply modifying the logic expressions that map time points to new event relations. The code is released at \url{https://github.com/AndrewZhe/A-Unified-Framework-for-ETRE}",
}
| Event temporal relation extraction (ETRE) is usually formulated as a multi-label classification task, where each type of relation is simply treated as a one-hot label. This formulation ignores the meaning of relations and wipes out their intrinsic dependency. After examining the relation definitions in various ETRE tasks, we observe that all relations can be interpreted using the start and end time points of events. For example, relation \textit{Includes} could be interpreted as event 1 starting no later than event 2 and ending no earlier than event 2. In this paper, we propose a unified event temporal relation extraction framework, which transforms temporal relations into logical expressions of time points and completes the ETRE by predicting the relations between certain time point pairs. Experiments on TB-Dense and MATRES show significant improvements over a strong baseline and outperform the state-of-the-art model by 0.3{\%} on both datasets. By representing all relations in a unified framework, we can leverage the relations with sufficient data to assist the learning of other relations, thus achieving stable improvement in low-data scenarios. When the relation definitions are changed, our method can quickly adapt to the new ones by simply modifying the logic expressions that map time points to new event relations. The code is released at \url{https://github.com/AndrewZhe/A-Unified-Framework-for-ETRE} | [
"Huang, Quzhe",
"Hu, Yutong",
"Zhu, Shengqi",
"Feng, Yansong",
"Liu, Chang",
"Zhao, Dongyan"
] | More than Classification: A Unified Framework for Event Temporal Relation Extraction | acl-long.536 | Poster | 2305.17607 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.537.bib | https://aclanthology.org/2023.acl-long.537/ | @inproceedings{ye-etal-2023-multi,
title = "Multi-Source Test-Time Adaptation as Dueling Bandits for Extractive Question Answering",
author = "Ye, Hai and
Xie, Qizhe and
Ng, Hwee Tou",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.537",
doi = "10.18653/v1/2023.acl-long.537",
pages = "9647--9660",
abstract = "In this work, we study multi-source test-time model adaptation from user feedback, where $K$ distinct models are established for adaptation. To allow efficient adaptation, we cast the problem as a stochastic decision-making process, aiming to determine the best adapted model after adaptation. We discuss two frameworks: multi-armed bandit learning and multi-armed dueling bandits. Compared to multi-armed bandit learning, the dueling framework allows pairwise collaboration among $K$ models, which is solved by a novel method named Co-UCB proposed in this work. Experiments on six datasets of extractive question answering (QA) show that the dueling framework using Co-UCB is more effective than other strong baselines for our studied problem.",
}
| In this work, we study multi-source test-time model adaptation from user feedback, where $K$ distinct models are established for adaptation. To allow efficient adaptation, we cast the problem as a stochastic decision-making process, aiming to determine the best adapted model after adaptation. We discuss two frameworks: multi-armed bandit learning and multi-armed dueling bandits. Compared to multi-armed bandit learning, the dueling framework allows pairwise collaboration among $K$ models, which is solved by a novel method named Co-UCB proposed in this work. Experiments on six datasets of extractive question answering (QA) show that the dueling framework using Co-UCB is more effective than other strong baselines for our studied problem. | [
"Ye, Hai",
"Xie, Qizhe",
"Ng, Hwee Tou"
] | Multi-Source Test-Time Adaptation as Dueling Bandits for Extractive Question Answering | acl-long.537 | Oral | 2306.06779 | [
"https://github.com/oceanypt/multi-source-tta"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.538.bib | https://aclanthology.org/2023.acl-long.538/ | @inproceedings{mou-etal-2023-decoupling,
title = "Decoupling Pseudo Label Disambiguation and Representation Learning for Generalized Intent Discovery",
author = "Mou, Yutao and
Song, Xiaoshuai and
He, Keqing and
Zeng, Chen and
Wang, Pei and
Wang, Jingang and
Xian, Yunsen and
Xu, Weiran",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.538",
doi = "10.18653/v1/2023.acl-long.538",
pages = "9661--9675",
abstract = "Generalized intent discovery aims to extend a closed-set in-domain intent classifier to an open-world intent set including in-domain and out-of-domain intents. The key challenges lie in pseudo label disambiguation and representation learning. Previous methods suffer from a coupling of pseudo label disambiguation and representation learning, that is, the reliability of pseudo labels relies on representation learning, and representation learning is restricted by pseudo labels in turn. In this paper, we propose a decoupled prototype learning framework (DPL) to decouple pseudo label disambiguation and representation learning. Specifically, we firstly introduce prototypical contrastive representation learning (PCL) to get discriminative representations. And then we adopt a prototype-based label disambiguation method (PLD) to obtain pseudo labels. We theoretically prove that PCL and PLD work in a collaborative fashion and facilitate pseudo label disambiguation. Experiments and analysis on three benchmark datasets show the effectiveness of our method.",
}
| Generalized intent discovery aims to extend a closed-set in-domain intent classifier to an open-world intent set including in-domain and out-of-domain intents. The key challenges lie in pseudo label disambiguation and representation learning. Previous methods suffer from a coupling of pseudo label disambiguation and representation learning, that is, the reliability of pseudo labels relies on representation learning, and representation learning is restricted by pseudo labels in turn. In this paper, we propose a decoupled prototype learning framework (DPL) to decouple pseudo label disambiguation and representation learning. Specifically, we firstly introduce prototypical contrastive representation learning (PCL) to get discriminative representations. And then we adopt a prototype-based label disambiguation method (PLD) to obtain pseudo labels. We theoretically prove that PCL and PLD work in a collaborative fashion and facilitate pseudo label disambiguation. Experiments and analysis on three benchmark datasets show the effectiveness of our method. | [
"Mou, Yutao",
"Song, Xiaoshuai",
"He, Keqing",
"Zeng, Chen",
"Wang, Pei",
"Wang, Jingang",
"Xian, Yunsen",
"Xu, Weiran"
] | Decoupling Pseudo Label Disambiguation and Representation Learning for Generalized Intent Discovery | acl-long.538 | Poster | 2305.17699 | [
"https://github.com/songxiaoshuai/dpl"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.539.bib | https://aclanthology.org/2023.acl-long.539/ | @inproceedings{ke-etal-2023-decompeval,
title = "{D}ecomp{E}val: Evaluating Generated Texts as Unsupervised Decomposed Question Answering",
author = "Ke, Pei and
Huang, Fei and
Mi, Fei and
Wang, Yasheng and
Liu, Qun and
Zhu, Xiaoyan and
Huang, Minlie",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.539",
doi = "10.18653/v1/2023.acl-long.539",
pages = "9676--9691",
abstract = "Existing evaluation metrics for natural language generation (NLG) tasks face the challenges on generalization ability and interpretability. Specifically, most of the well-performed metrics are required to train on evaluation datasets of specific NLG tasks and evaluation dimensions, which may cause over-fitting to task-specific datasets. Furthermore, existing metrics only provide an evaluation score for each dimension without revealing the evidence to interpret how this score is obtained. To deal with these challenges, we propose a simple yet effective metric called DecompEval. This metric formulates NLG evaluation as an instruction-style question answering task and utilizes instruction-tuned pre-trained language models (PLMs) without training on evaluation datasets, aiming to enhance the generalization ability. To make the evaluation process more interpretable, we decompose our devised instruction-style question about the quality of generated texts into the subquestions that measure the quality of each sentence. The subquestions with their answers generated by PLMs are then recomposed as evidence to obtain the evaluation result. Experimental results show that DecompEval achieves state-of-the-art performance in untrained metrics for evaluating text summarization and dialogue generation, which also exhibits strong dimension-level / task-level generalization ability and interpretability.",
}
| Existing evaluation metrics for natural language generation (NLG) tasks face the challenges on generalization ability and interpretability. Specifically, most of the well-performed metrics are required to train on evaluation datasets of specific NLG tasks and evaluation dimensions, which may cause over-fitting to task-specific datasets. Furthermore, existing metrics only provide an evaluation score for each dimension without revealing the evidence to interpret how this score is obtained. To deal with these challenges, we propose a simple yet effective metric called DecompEval. This metric formulates NLG evaluation as an instruction-style question answering task and utilizes instruction-tuned pre-trained language models (PLMs) without training on evaluation datasets, aiming to enhance the generalization ability. To make the evaluation process more interpretable, we decompose our devised instruction-style question about the quality of generated texts into the subquestions that measure the quality of each sentence. The subquestions with their answers generated by PLMs are then recomposed as evidence to obtain the evaluation result. Experimental results show that DecompEval achieves state-of-the-art performance in untrained metrics for evaluating text summarization and dialogue generation, which also exhibits strong dimension-level / task-level generalization ability and interpretability. | [
"Ke, Pei",
"Huang, Fei",
"Mi, Fei",
"Wang, Yasheng",
"Liu, Qun",
"Zhu, Xiaoyan",
"Huang, Minlie"
] | DecompEval: Evaluating Generated Texts as Unsupervised Decomposed Question Answering | acl-long.539 | Poster | 2307.06869 | [
"https://github.com/kepei1106/decompeval"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.540.bib | https://aclanthology.org/2023.acl-long.540/ | @inproceedings{sun-etal-2023-backdooring,
title = "Backdooring Neural Code Search",
author = "Sun, Weisong and
Chen, Yuchen and
Tao, Guanhong and
Fang, Chunrong and
Zhang, Xiangyu and
Zhang, Quanjun and
Luo, Bin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.540",
doi = "10.18653/v1/2023.acl-long.540",
pages = "9692--9708",
abstract = "Reusing off-the-shelf code snippets from online repositories is a common practice, which significantly enhances the productivity of software developers. To find desired code snippets, developers resort to code search engines through natural language queries. Neural code search models are hence behind many such engines. These models are based on deep learning and gain substantial attention due to their impressive performance. However, the security aspect of these models is rarely studied. Particularly, an adversary can inject a backdoor in neural code search models, which return buggy or even vulnerable code with security/privacy issues. This may impact the downstream software (e.g., stock trading systems and autonomous driving) and cause financial loss and/or life-threatening incidents. In this paper, we demonstrate such attacks are feasible and can be quite stealthy. By simply modifying one variable/function name, the attacker can make buggy/vulnerable code rank in the top 11{\%}. Our attack BADCODE features a special trigger generation and injection procedure, making the attack more effective and stealthy. The evaluation is conducted on two neural code search models and the results show our attack outperforms baselines by 60{\%}. Our user study demonstrates that our attack is more stealthy than the baseline by two times based on the F1 score.",
}
| Reusing off-the-shelf code snippets from online repositories is a common practice, which significantly enhances the productivity of software developers. To find desired code snippets, developers resort to code search engines through natural language queries. Neural code search models are hence behind many such engines. These models are based on deep learning and gain substantial attention due to their impressive performance. However, the security aspect of these models is rarely studied. Particularly, an adversary can inject a backdoor in neural code search models, which return buggy or even vulnerable code with security/privacy issues. This may impact the downstream software (e.g., stock trading systems and autonomous driving) and cause financial loss and/or life-threatening incidents. In this paper, we demonstrate such attacks are feasible and can be quite stealthy. By simply modifying one variable/function name, the attacker can make buggy/vulnerable code rank in the top 11{\%}. Our attack BADCODE features a special trigger generation and injection procedure, making the attack more effective and stealthy. The evaluation is conducted on two neural code search models and the results show our attack outperforms baselines by 60{\%}. Our user study demonstrates that our attack is more stealthy than the baseline by two times based on the F1 score. | [
"Sun, Weisong",
"Chen, Yuchen",
"Tao, Guanhong",
"Fang, Chunrong",
"Zhang, Xiangyu",
"Zhang, Quanjun",
"Luo, Bin"
] | Backdooring Neural Code Search | acl-long.540 | Poster | 2305.17506 | [
"https://github.com/wssun/badcode"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.541.bib | https://aclanthology.org/2023.acl-long.541/ | @inproceedings{potluri-etal-2023-concise,
title = "Concise Answers to Complex Questions: Summarization of Long-form Answers",
author = "Potluri, Abhilash and
Xu, Fangyuan and
Choi, Eunsol",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.541",
doi = "10.18653/v1/2023.acl-long.541",
pages = "9709--9728",
abstract = "Long-form question answering systems provide rich information by presenting paragraph-level answers, often containing optional background or auxiliary information. While such comprehensive answers are helpful, not all information is required to answer the question (e.g. users with domain knowledge do not need an explanation of background). Can we provide a concise version of the answer by summarizing it, while still addressing the question? We conduct a user study on summarized answers generated from state-of-the-art models and our newly proposed extract-and-decontextualize approach. We find a large proportion of long-form answers (over 90{\%}) in the ELI5 domain can be adequately summarized by at least one system, while complex and implicit answers are challenging to compress. We observe that decontextualization improves the quality of the extractive summary, exemplifying its potential in the summarization task. To promote future work, we provide an extractive summarization dataset covering 1K long-form answers and our user study annotations. Together, we present the first study on summarizing long-form answers, taking a step forward for QA agents that can provide answers at multiple granularities.",
}
| Long-form question answering systems provide rich information by presenting paragraph-level answers, often containing optional background or auxiliary information. While such comprehensive answers are helpful, not all information is required to answer the question (e.g. users with domain knowledge do not need an explanation of background). Can we provide a concise version of the answer by summarizing it, while still addressing the question? We conduct a user study on summarized answers generated from state-of-the-art models and our newly proposed extract-and-decontextualize approach. We find a large proportion of long-form answers (over 90{\%}) in the ELI5 domain can be adequately summarized by at least one system, while complex and implicit answers are challenging to compress. We observe that decontextualization improves the quality of the extractive summary, exemplifying its potential in the summarization task. To promote future work, we provide an extractive summarization dataset covering 1K long-form answers and our user study annotations. Together, we present the first study on summarizing long-form answers, taking a step forward for QA agents that can provide answers at multiple granularities. | [
"Potluri, Abhilash",
"Xu, Fangyuan",
"Choi, Eunsol"
] | Concise Answers to Complex Questions: Summarization of Long-form Answers | acl-long.541 | Poster | 2305.19271 | [
"https://github.com/acpotluri/lfqa_summary"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.542.bib | https://aclanthology.org/2023.acl-long.542/ | @inproceedings{liu-etal-2023-towards-better,
title = "Towards Better Entity Linking with Multi-View Enhanced Distillation",
author = "Liu, Yi and
Tian, Yuan and
Lian, Jianxun and
Wang, Xinlong and
Cao, Yanan and
Fang, Fang and
Zhang, Wen and
Huang, Haizhen and
Deng, Weiwei and
Zhang, Qi",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.542",
doi = "10.18653/v1/2023.acl-long.542",
pages = "9729--9743",
abstract = "Dense retrieval is widely used for entity linking to retrieve entities from large-scale knowledge bases. Mainstream techniques are based on a dual-encoder framework, which encodes mentions and entities independently and calculates their relevances via rough interaction metrics, resulting in difficulty in explicitly modeling multiple mention-relevant parts within entities to match divergent mentions. Aiming at learning entity representations that can match divergent mentions, this paper proposes a Multi-View Enhanced Distillation (MVD) framework, which can effectively transfer knowledge of multiple fine-grained and mention-relevant parts within entities from cross-encoders to dual-encoders. Each entity is split into multiple views to avoid irrelevant information being over-squashed into the mention-relevant view. We further design cross-alignment and self-alignment mechanisms for this framework to facilitate fine-grained knowledge distillation from the teacher model to the student model. Meanwhile, we reserve a global-view that embeds the entity as a whole to prevent dispersal of uniform information. Experiments show our method achieves state-of-the-art performance on several entity linking benchmarks.",
}
| Dense retrieval is widely used for entity linking to retrieve entities from large-scale knowledge bases. Mainstream techniques are based on a dual-encoder framework, which encodes mentions and entities independently and calculates their relevances via rough interaction metrics, resulting in difficulty in explicitly modeling multiple mention-relevant parts within entities to match divergent mentions. Aiming at learning entity representations that can match divergent mentions, this paper proposes a Multi-View Enhanced Distillation (MVD) framework, which can effectively transfer knowledge of multiple fine-grained and mention-relevant parts within entities from cross-encoders to dual-encoders. Each entity is split into multiple views to avoid irrelevant information being over-squashed into the mention-relevant view. We further design cross-alignment and self-alignment mechanisms for this framework to facilitate fine-grained knowledge distillation from the teacher model to the student model. Meanwhile, we reserve a global-view that embeds the entity as a whole to prevent dispersal of uniform information. Experiments show our method achieves state-of-the-art performance on several entity linking benchmarks. | [
"Liu, Yi",
"Tian, Yuan",
"Lian, Jianxun",
"Wang, Xinlong",
"Cao, Yanan",
"Fang, Fang",
"Zhang, Wen",
"Huang, Haizhen",
"Deng, Weiwei",
"Zhang, Qi"
] | Towards Better Entity Linking with Multi-View Enhanced Distillation | acl-long.542 | Poster | 2305.17371 | [
"https://github.com/noen61/mvd"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.543.bib | https://aclanthology.org/2023.acl-long.543/ | @inproceedings{du-etal-2023-measure,
title = "A Measure-Theoretic Characterization of Tight Language Models",
author = "Du, Li and
Torroba Hennigen, Lucas and
Pimentel, Tiago and
Meister, Clara and
Eisner, Jason and
Cotterell, Ryan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.543",
doi = "10.18653/v1/2023.acl-long.543",
pages = "9744--9770",
abstract = "Language modeling, a central task in natural language processing, involves estimating a probability distribution over strings. In most cases, the estimated distribution sums to 1 over all finite strings. However, in some pathological cases, probability mass can {``}leak{''} onto the set of infinite sequences. In order to characterize the notion of leakage more precisely, this paper offers a measure-theoretic treatment of language modeling. We prove that many popular language model families are in fact tight, meaning that they will not leak in this sense. We also generalize characterizations of tightness proposed in previous works.",
}
| Language modeling, a central task in natural language processing, involves estimating a probability distribution over strings. In most cases, the estimated distribution sums to 1 over all finite strings. However, in some pathological cases, probability mass can {``}leak{''} onto the set of infinite sequences. In order to characterize the notion of leakage more precisely, this paper offers a measure-theoretic treatment of language modeling. We prove that many popular language model families are in fact tight, meaning that they will not leak in this sense. We also generalize characterizations of tightness proposed in previous works. | [
"Du, Li",
"Torroba Hennigen, Lucas",
"Pimentel, Tiago",
"Meister, Clara",
"Eisner, Jason",
"Cotterell, Ryan"
] | A Measure-Theoretic Characterization of Tight Language Models | acl-long.543 | Poster | 2212.10502 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.544.bib | https://aclanthology.org/2023.acl-long.544/ | @inproceedings{zhu-etal-2023-paed,
title = "{PAED}: Zero-Shot Persona Attribute Extraction in Dialogues",
author = "Zhu, Luyao and
Li, Wei and
Mao, Rui and
Pandelea, Vlad and
Cambria, Erik",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.544",
doi = "10.18653/v1/2023.acl-long.544",
pages = "9771--9787",
abstract = "Persona attribute extraction is critical for personalized human-computer interaction. Dialogue is an important medium that communicates and delivers persona information. Although there is a public dataset for triplet-based persona attribute extraction from conversations, its automatically generated labels present many issues, including unspecific relations and inconsistent annotations. We fix such issues by leveraging more reliable text-label matching criteria to generate high-quality data for persona attribute extraction. We also propose a contrastive learning- and generation-based model with a novel hard negative sampling strategy for generalized zero-shot persona attribute extraction. We benchmark our model with state-of-the-art baselines on our dataset and a public dataset, showing outstanding accuracy gains. Our sampling strategy also exceeds others by a large margin in persona attribute extraction.",
}
| Persona attribute extraction is critical for personalized human-computer interaction. Dialogue is an important medium that communicates and delivers persona information. Although there is a public dataset for triplet-based persona attribute extraction from conversations, its automatically generated labels present many issues, including unspecific relations and inconsistent annotations. We fix such issues by leveraging more reliable text-label matching criteria to generate high-quality data for persona attribute extraction. We also propose a contrastive learning- and generation-based model with a novel hard negative sampling strategy for generalized zero-shot persona attribute extraction. We benchmark our model with state-of-the-art baselines on our dataset and a public dataset, showing outstanding accuracy gains. Our sampling strategy also exceeds others by a large margin in persona attribute extraction. | [
"Zhu, Luyao",
"Li, Wei",
"Mao, Rui",
"P",
"elea, Vlad",
"Cambria, Erik"
] | PAED: Zero-Shot Persona Attribute Extraction in Dialogues | acl-long.544 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.545.bib | https://aclanthology.org/2023.acl-long.545/ | @inproceedings{kong-etal-2023-promptrank,
title = "{P}rompt{R}ank: Unsupervised Keyphrase Extraction Using Prompt",
author = "Kong, Aobo and
Zhao, Shiwan and
Chen, Hao and
Li, Qicheng and
Qin, Yong and
Sun, Ruiqi and
Bai, Xiaoyan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.545",
doi = "10.18653/v1/2023.acl-long.545",
pages = "9788--9801",
abstract = "The keyphrase extraction task refers to the automatic selection of phrases from a given document to summarize its core content. State-of-the-art (SOTA) performance has recently been achieved by embedding-based algorithms, which rank candidates according to how similar their embeddings are to document embeddings. However, such solutions either struggle with the document and candidate length discrepancies or fail to fully utilize the pre-trained language model (PLM) without further fine-tuning. To this end, in this paper, we propose a simple yet effective unsupervised approach, PromptRank, based on the PLM with an encoder-decoder architecture. Specifically, PromptRank feeds the document into the encoder and calculates the probability of generating the candidate with a designed prompt by the decoder. We extensively evaluate the proposed PromptRank on six widely used benchmarks. PromptRank outperforms the SOTA approach MDERank, improving the F1 score relatively by 34.18{\%}, 24.87{\%}, and 17.57{\%} for 5, 10, and 15 returned results, respectively. This demonstrates the great potential of using prompt for unsupervised keyphrase extraction. We release our code at \url{https://github.com/HLT-NLP/PromptRank}.",
}
| The keyphrase extraction task refers to the automatic selection of phrases from a given document to summarize its core content. State-of-the-art (SOTA) performance has recently been achieved by embedding-based algorithms, which rank candidates according to how similar their embeddings are to document embeddings. However, such solutions either struggle with the document and candidate length discrepancies or fail to fully utilize the pre-trained language model (PLM) without further fine-tuning. To this end, in this paper, we propose a simple yet effective unsupervised approach, PromptRank, based on the PLM with an encoder-decoder architecture. Specifically, PromptRank feeds the document into the encoder and calculates the probability of generating the candidate with a designed prompt by the decoder. We extensively evaluate the proposed PromptRank on six widely used benchmarks. PromptRank outperforms the SOTA approach MDERank, improving the F1 score relatively by 34.18{\%}, 24.87{\%}, and 17.57{\%} for 5, 10, and 15 returned results, respectively. This demonstrates the great potential of using prompt for unsupervised keyphrase extraction. We release our code at \url{https://github.com/HLT-NLP/PromptRank}. | [
"Kong, Aobo",
"Zhao, Shiwan",
"Chen, Hao",
"Li, Qicheng",
"Qin, Yong",
"Sun, Ruiqi",
"Bai, Xiaoyan"
] | PromptRank: Unsupervised Keyphrase Extraction Using Prompt | acl-long.545 | Poster | 2305.04490 | [
"https://github.com/hlt-nlp/promptrank"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.546.bib | https://aclanthology.org/2023.acl-long.546/ | @inproceedings{mallen-etal-2023-trust,
title = "When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories",
author = "Mallen, Alex and
Asai, Akari and
Zhong, Victor and
Das, Rajarshi and
Khashabi, Daniel and
Hajishirzi, Hannaneh",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.546",
doi = "10.18653/v1/2023.acl-long.546",
pages = "9802--9822",
abstract = "Despite their impressive performance on diverse tasks, large language models (LMs) still struggle with tasks requiring rich world knowledge, implying the difficulty of encoding a wealth of world knowledge in their parameters. This paper aims to understand LMs{'} strengths and limitations in memorizing factual knowledge, by conducting large-scale knowledge probing experiments on two open-domain entity-centric QA datasets: PopQA, our new dataset with 14k questions about long-tail entities, and EntityQuestions, a widely used open-domain QA dataset. We find that LMs struggle with less popular factual knowledge, and that retrieval augmentation helps significantly in these cases. Scaling, on the other hand, mainly improves memorization of popular knowledge, and fails to appreciably improve memorization of factual knowledge in the tail. Based on those findings, we devise a new method for retrieval-augmentation that improves performance and reduces inference costs by only retrieving non-parametric memories when necessary.",
}
| Despite their impressive performance on diverse tasks, large language models (LMs) still struggle with tasks requiring rich world knowledge, implying the difficulty of encoding a wealth of world knowledge in their parameters. This paper aims to understand LMs{'} strengths and limitations in memorizing factual knowledge, by conducting large-scale knowledge probing experiments on two open-domain entity-centric QA datasets: PopQA, our new dataset with 14k questions about long-tail entities, and EntityQuestions, a widely used open-domain QA dataset. We find that LMs struggle with less popular factual knowledge, and that retrieval augmentation helps significantly in these cases. Scaling, on the other hand, mainly improves memorization of popular knowledge, and fails to appreciably improve memorization of factual knowledge in the tail. Based on those findings, we devise a new method for retrieval-augmentation that improves performance and reduces inference costs by only retrieving non-parametric memories when necessary. | [
"Mallen, Alex",
"Asai, Akari",
"Zhong, Victor",
"Das, Rajarshi",
"Khashabi, Daniel",
"Hajishirzi, Hannaneh"
] | When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories | acl-long.546 | Oral | 2212.10511 | [
"https://github.com/alextmallen/adaptive-retrieval"
] | https://huggingface.co/papers/2212.10511 | 2 | 1 | 0 | 6 | 1 | [
"princeton-nlp/Llama-3-8B-ProLong-64k-Instruct",
"QuantFactory/Llama-3-8B-ProLong-64k-Instruct-GGUF",
"princeton-nlp/Llama-3-8B-ProLong-64k-Base",
"RichardErkhov/princeton-nlp_-_Llama-3-8B-ProLong-64k-Base-gguf"
] | [] | [] |
https://aclanthology.org/2023.acl-long.547.bib | https://aclanthology.org/2023.acl-long.547/ | @inproceedings{kim-etal-2023-infoverse,
title = "info{V}erse: A Universal Framework for Dataset Characterization with Multidimensional Meta-information",
author = "Kim, Jaehyung and
Kim, Yekyung and
de Langis, Karin and
Shin, Jinwoo and
Kang, Dongyeop",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.547",
doi = "10.18653/v1/2023.acl-long.547",
pages = "9823--9850",
abstract = "The success of NLP systems often relies on the availability of large, high-quality datasets. However, not all samples in these datasets are equally valuable for learning, as some may be redundant or noisy. Several methods for characterizing datasets based on model-driven meta-information (e.g., model{'}s confidence) have been developed, but the relationship and complementary effects of these methods have received less attention. In this paper, we introduce infoVerse, a universal framework for dataset characterization, which provides a new feature space that effectively captures multidimensional characteristics of datasets by incorporating various model-driven meta-information. infoVerse reveals distinctive regions of the dataset that are not apparent in the original semantic space, hence guiding users (or models) in identifying which samples to focus on for exploration, assessment, or annotation. Additionally, we propose a novel sampling method on infoVerse to select a set of data points that maximizes informativeness. In three real-world applications (data pruning, active learning, and data annotation), the samples chosen on infoVerse space consistently outperform strong baselines in all applications. Our code and demo are publicly available.",
}
| The success of NLP systems often relies on the availability of large, high-quality datasets. However, not all samples in these datasets are equally valuable for learning, as some may be redundant or noisy. Several methods for characterizing datasets based on model-driven meta-information (e.g., model{'}s confidence) have been developed, but the relationship and complementary effects of these methods have received less attention. In this paper, we introduce infoVerse, a universal framework for dataset characterization, which provides a new feature space that effectively captures multidimensional characteristics of datasets by incorporating various model-driven meta-information. infoVerse reveals distinctive regions of the dataset that are not apparent in the original semantic space, hence guiding users (or models) in identifying which samples to focus on for exploration, assessment, or annotation. Additionally, we propose a novel sampling method on infoVerse to select a set of data points that maximizes informativeness. In three real-world applications (data pruning, active learning, and data annotation), the samples chosen on infoVerse space consistently outperform strong baselines in all applications. Our code and demo are publicly available. | [
"Kim, Jaehyung",
"Kim, Yekyung",
"de Langis, Karin",
"Shin, Jinwoo",
"Kang, Dongyeop"
] | infoVerse: A Universal Framework for Dataset Characterization with Multidimensional Meta-information | acl-long.547 | Poster | 2305.19344 | [
"https://github.com/minnesotanlp/infoverse"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.548.bib | https://aclanthology.org/2023.acl-long.548/ | @inproceedings{jha-etal-2023-seegull,
title = "{S}ee{GULL}: A Stereotype Benchmark with Broad Geo-Cultural Coverage Leveraging Generative Models",
author = "Jha, Akshita and
Mostafazadeh Davani, Aida and
Reddy, Chandan K and
Dave, Shachi and
Prabhakaran, Vinodkumar and
Dev, Sunipa",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.548",
doi = "10.18653/v1/2023.acl-long.548",
pages = "9851--9870",
abstract = "Stereotype benchmark datasets are crucial to detect and mitigate social stereotypes about groups of people in NLP models. However, existing datasets are limited in size and coverage, and are largely restricted to stereotypes prevalent in the Western society. This is especially problematic as language technologies gain hold across the globe. To address this gap, we present SeeGULL, a broad-coverage stereotype dataset, built by utilizing generative capabilities of large language models such as PaLM, and GPT-3, and leveraging a globally diverse rater pool to validate the prevalence of those stereotypes in society. SeeGULL is in English, and contains stereotypes about identity groups spanning 178 countries across 8 different geo-political regions across 6 continents, as well as state-level identities within the US and India. We also include fine-grained offensiveness scores for different stereotypes and demonstrate their global disparities. Furthermore, we include comparative annotations about the same groups by annotators living in the region vs. those that are based in North America, and demonstrate that within-region stereotypes about groups differ from those prevalent in North America.",
}
| Stereotype benchmark datasets are crucial to detect and mitigate social stereotypes about groups of people in NLP models. However, existing datasets are limited in size and coverage, and are largely restricted to stereotypes prevalent in the Western society. This is especially problematic as language technologies gain hold across the globe. To address this gap, we present SeeGULL, a broad-coverage stereotype dataset, built by utilizing generative capabilities of large language models such as PaLM, and GPT-3, and leveraging a globally diverse rater pool to validate the prevalence of those stereotypes in society. SeeGULL is in English, and contains stereotypes about identity groups spanning 178 countries across 8 different geo-political regions across 6 continents, as well as state-level identities within the US and India. We also include fine-grained offensiveness scores for different stereotypes and demonstrate their global disparities. Furthermore, we include comparative annotations about the same groups by annotators living in the region vs. those that are based in North America, and demonstrate that within-region stereotypes about groups differ from those prevalent in North America. | [
"Jha, Akshita",
"Mostafazadeh Davani, Aida",
"Reddy, Ch",
"an K",
"Dave, Shachi",
"Prabhakaran, Vinodkumar",
"Dev, Sunipa"
] | SeeGULL: A Stereotype Benchmark with Broad Geo-Cultural Coverage Leveraging Generative Models | acl-long.548 | Poster | 2305.11840 | [
"https://github.com/google-research-datasets/seegull"
] | https://huggingface.co/papers/2305.11840 | 3 | 1 | 0 | 6 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.549.bib | https://aclanthology.org/2023.acl-long.549/ | @inproceedings{wang-etal-2023-automated,
title = "Automated Metrics for Medical Multi-Document Summarization Disagree with Human Evaluations",
author = "Wang, Lucy Lu and
Otmakhova, Yulia and
DeYoung, Jay and
Truong, Thinh Hung and
Kuehl, Bailey and
Bransom, Erin and
Wallace, Byron",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.549",
doi = "10.18653/v1/2023.acl-long.549",
pages = "9871--9889",
abstract = "Evaluating multi-document summarization (MDS) quality is difficult. This is especially true in the case of MDS for biomedical literature reviews, where models must synthesize contradicting evidence reported across different documents. Prior work has shown that rather than performing the task, models may exploit shortcuts that are difficult to detect using standard n-gram similarity metrics such as ROUGE. Better automated evaluation metrics are needed, but few resources exist to assess metrics when they are proposed. Therefore, we introduce a dataset of human-assessed summary quality facets and pairwise preferences to encourage and support the development of better automated evaluation methods for literature review MDS. We take advantage of community submissions to the Multi-document Summarization for Literature Review (MSLR) shared task to compile a diverse and representative sample of generated summaries. We analyze how automated summarization evaluation metrics correlate with lexical features of generated summaries, to other automated metrics including several we propose in this work, and to aspects of human-assessed summary quality. We find that not only do automated metrics fail to capture aspects of quality as assessed by humans, in many cases the system rankings produced by these metrics are anti-correlated with rankings according to human annotators.",
}
| Evaluating multi-document summarization (MDS) quality is difficult. This is especially true in the case of MDS for biomedical literature reviews, where models must synthesize contradicting evidence reported across different documents. Prior work has shown that rather than performing the task, models may exploit shortcuts that are difficult to detect using standard n-gram similarity metrics such as ROUGE. Better automated evaluation metrics are needed, but few resources exist to assess metrics when they are proposed. Therefore, we introduce a dataset of human-assessed summary quality facets and pairwise preferences to encourage and support the development of better automated evaluation methods for literature review MDS. We take advantage of community submissions to the Multi-document Summarization for Literature Review (MSLR) shared task to compile a diverse and representative sample of generated summaries. We analyze how automated summarization evaluation metrics correlate with lexical features of generated summaries, to other automated metrics including several we propose in this work, and to aspects of human-assessed summary quality. We find that not only do automated metrics fail to capture aspects of quality as assessed by humans, in many cases the system rankings produced by these metrics are anti-correlated with rankings according to human annotators. | [
"Wang, Lucy Lu",
"Otmakhova, Yulia",
"DeYoung, Jay",
"Truong, Thinh Hung",
"Kuehl, Bailey",
"Bransom, Erin",
"Wallace, Byron"
] | Automated Metrics for Medical Multi-Document Summarization Disagree with Human Evaluations | acl-long.549 | Poster | 2305.13693 | [
"https://github.com/allenai/mslr-annotated-dataset"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.550.bib | https://aclanthology.org/2023.acl-long.550/ | @inproceedings{chen-etal-2023-say,
title = "Say What You Mean! Large Language Models Speak Too Positively about Negative Commonsense Knowledge",
author = "Chen, Jiangjie and
Shi, Wei and
Fu, Ziquan and
Cheng, Sijie and
Li, Lei and
Xiao, Yanghua",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.550",
doi = "10.18653/v1/2023.acl-long.550",
pages = "9890--9908",
abstract = "Large language models (LLMs) have been widely studied for their ability to store and utilize positive knowledge. However, negative knowledge, such as {``}lions don{'}t live in the ocean{''}, is also ubiquitous in the world but rarely mentioned explicitly in text. What do LLMs know about negative knowledge?This work examines the ability of LLMs on negative commonsense knowledge. We design a constrained keywords-to-sentence generation task (CG) and a Boolean question answering task (QA) to probe LLMs.Our experiments reveal that LLMs frequently fail to generate valid sentences grounded in negative commonsense knowledge, yet they can correctly answer polar yes-or-no questions. We term this phenomenon the belief conflict of LLMs.Our further analysis shows that statistical shortcuts and negation reporting bias from language modeling pre-training cause this conflict.",
}
| Large language models (LLMs) have been widely studied for their ability to store and utilize positive knowledge. However, negative knowledge, such as {``}lions don{'}t live in the ocean{''}, is also ubiquitous in the world but rarely mentioned explicitly in text. What do LLMs know about negative knowledge?This work examines the ability of LLMs on negative commonsense knowledge. We design a constrained keywords-to-sentence generation task (CG) and a Boolean question answering task (QA) to probe LLMs.Our experiments reveal that LLMs frequently fail to generate valid sentences grounded in negative commonsense knowledge, yet they can correctly answer polar yes-or-no questions. We term this phenomenon the belief conflict of LLMs.Our further analysis shows that statistical shortcuts and negation reporting bias from language modeling pre-training cause this conflict. | [
"Chen, Jiangjie",
"Shi, Wei",
"Fu, Ziquan",
"Cheng, Sijie",
"Li, Lei",
"Xiao, Yanghua"
] | Say What You Mean! Large Language Models Speak Too Positively about Negative Commonsense Knowledge | acl-long.550 | Poster | 2305.05976 | [
"https://github.com/jiangjiechen/uncommongen"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.551.bib | https://aclanthology.org/2023.acl-long.551/ | @inproceedings{lin-etal-2023-inner,
title = "An Inner Table Retriever for Robust Table Question Answering",
author = "Lin, Weizhe and
Blloshmi, Rexhina and
Byrne, Bill and
de Gispert, Adria and
Iglesias, Gonzalo",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.551",
doi = "10.18653/v1/2023.acl-long.551",
pages = "9909--9926",
abstract = "Recent years have witnessed the thriving of pretrained Transformer-based language models for understanding semi-structured tables, with several applications, such as Table Question Answering (TableQA).These models are typically trained on joint tables and surrounding natural language text, by linearizing table content into sequences comprising special tokens and cell information. This yields very long sequences which increase system inefficiency, and moreover, simply truncating long sequences results in information loss for downstream tasks. We propose Inner Table Retriever (ITR), a general-purpose approach for handling long tables in TableQA that extracts sub-tables to preserve the most relevant information for a question. We show that ITR can be easily integrated into existing systems to improve their accuracy with up to 1.3-4.8{\%} and achieve state-of-the-art results in two benchmarks, i.e., 63.4{\%} in WikiTableQuestions and 92.1{\%} in WikiSQL. Additionally, we show that ITR makes TableQA systems more robust to reduced model capacity and to different ordering of columns and rows. We make our code available at: \url{https://github.com/amazon-science/robust-tableqa}.",
}
| Recent years have witnessed the thriving of pretrained Transformer-based language models for understanding semi-structured tables, with several applications, such as Table Question Answering (TableQA).These models are typically trained on joint tables and surrounding natural language text, by linearizing table content into sequences comprising special tokens and cell information. This yields very long sequences which increase system inefficiency, and moreover, simply truncating long sequences results in information loss for downstream tasks. We propose Inner Table Retriever (ITR), a general-purpose approach for handling long tables in TableQA that extracts sub-tables to preserve the most relevant information for a question. We show that ITR can be easily integrated into existing systems to improve their accuracy with up to 1.3-4.8{\%} and achieve state-of-the-art results in two benchmarks, i.e., 63.4{\%} in WikiTableQuestions and 92.1{\%} in WikiSQL. Additionally, we show that ITR makes TableQA systems more robust to reduced model capacity and to different ordering of columns and rows. We make our code available at: \url{https://github.com/amazon-science/robust-tableqa}. | [
"Lin, Weizhe",
"Blloshmi, Rexhina",
"Byrne, Bill",
"de Gispert, Adria",
"Iglesias, Gonzalo"
] | An Inner Table Retriever for Robust Table Question Answering | acl-long.551 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.552.bib | https://aclanthology.org/2023.acl-long.552/ | @inproceedings{blinova-etal-2023-simsum,
title = "{SIMSUM}: Document-level Text Simplification via Simultaneous Summarization",
author = "Blinova, Sofia and
Zhou, Xinyu and
Jaggi, Martin and
Eickhoff, Carsten and
Bahrainian, Seyed Ali",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.552",
doi = "10.18653/v1/2023.acl-long.552",
pages = "9927--9944",
abstract = "Document-level text simplification is a specific type of simplification which involves simplifying documents consisting of several sentences by rewriting them into fewer or more sentences. In this paper, we propose a new two-stage framework SIMSUM for automated document-level text simplification. Our model is designed with explicit summarization and simplification models and guides the generation using the main keywords of a source text. In order to evaluate our new model, we use two existing benchmark datasets for simplification, namely D-Wikipedia and Wiki-Doc. We compare our model{'}s performance with state of the art and show that SIMSUM achieves top results on the D-Wikipedia dataset SARI (+1.20), D-SARI (+1.64), and FKGL (-0.35) scores, improving over the best baseline models. In order to evaluate the quality of the generated text, we analyze the outputs from different models qualitatively and demonstrate the merit of our new model. Our code and datasets are available.",
}
| Document-level text simplification is a specific type of simplification which involves simplifying documents consisting of several sentences by rewriting them into fewer or more sentences. In this paper, we propose a new two-stage framework SIMSUM for automated document-level text simplification. Our model is designed with explicit summarization and simplification models and guides the generation using the main keywords of a source text. In order to evaluate our new model, we use two existing benchmark datasets for simplification, namely D-Wikipedia and Wiki-Doc. We compare our model{'}s performance with state of the art and show that SIMSUM achieves top results on the D-Wikipedia dataset SARI (+1.20), D-SARI (+1.64), and FKGL (-0.35) scores, improving over the best baseline models. In order to evaluate the quality of the generated text, we analyze the outputs from different models qualitatively and demonstrate the merit of our new model. Our code and datasets are available. | [
"Blinova, Sofia",
"Zhou, Xinyu",
"Jaggi, Martin",
"Eickhoff, Carsten",
"Bahrainian, Seyed Ali"
] | SIMSUM: Document-level Text Simplification via Simultaneous Summarization | acl-long.552 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.553.bib | https://aclanthology.org/2023.acl-long.553/ | @inproceedings{zhou-etal-2023-simoap,
title = "{S}im{OAP}: Improve Coherence and Consistency in Persona-based Dialogue Generation via Over-sampling and Post-evaluation",
author = "Zhou, Junkai and
Pang, Liang and
Shen, Huawei and
Cheng, Xueqi",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.553",
doi = "10.18653/v1/2023.acl-long.553",
pages = "9945--9959",
abstract = "Language models trained on large-scale corpora can generate remarkably fluent results in open-domain dialogue. However, for the persona-based dialogue generation task, consistency and coherence are also key factors, which are great challenges for language models. Existing works mainly focus on valuable data filtering, model structure modifying, or objective function designing, while their improvements are limited and hard to generalize to all types of pre-trained language models. However, we find that language models can produce consistent and coherent responses if we consider enough generations. Thus, the problems lay in large-scale response generation and target response selection. In this work, a simple but effective two-stage SimOAP strategy is proposed, i.e., over-sampling and post-evaluation. The over-sampling stage takes large-scale responses from existing trained models efficiently via off-the-shelf distilling and compressing methods, and the post-evaluation stage selects a good response based on multiple well-designed evaluation metrics from large-scale candidates. Experimental results show that the proposed plug-in SimOAP strategy improves the backbone models and outperforms the baseline strategies in both automatic and human evaluations.",
}
| Language models trained on large-scale corpora can generate remarkably fluent results in open-domain dialogue. However, for the persona-based dialogue generation task, consistency and coherence are also key factors, which are great challenges for language models. Existing works mainly focus on valuable data filtering, model structure modifying, or objective function designing, while their improvements are limited and hard to generalize to all types of pre-trained language models. However, we find that language models can produce consistent and coherent responses if we consider enough generations. Thus, the problems lay in large-scale response generation and target response selection. In this work, a simple but effective two-stage SimOAP strategy is proposed, i.e., over-sampling and post-evaluation. The over-sampling stage takes large-scale responses from existing trained models efficiently via off-the-shelf distilling and compressing methods, and the post-evaluation stage selects a good response based on multiple well-designed evaluation metrics from large-scale candidates. Experimental results show that the proposed plug-in SimOAP strategy improves the backbone models and outperforms the baseline strategies in both automatic and human evaluations. | [
"Zhou, Junkai",
"Pang, Liang",
"Shen, Huawei",
"Cheng, Xueqi"
] | SimOAP: Improve Coherence and Consistency in Persona-based Dialogue Generation via Over-sampling and Post-evaluation | acl-long.553 | Poster | 2305.11130 | [
"https://github.com/934865517zjk/simoap"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.554.bib | https://aclanthology.org/2023.acl-long.554/ | @inproceedings{zheng-zhu-2023-natlogattack,
title = "{N}at{L}og{A}ttack: A Framework for Attacking Natural Language Inference Models with Natural Logic",
author = "Zheng, Zi{'}ou and
Zhu, Xiaodan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.554",
doi = "10.18653/v1/2023.acl-long.554",
pages = "9960--9976",
abstract = "Reasoning has been a central topic in artificial intelligence from the beginning. The recent progress made on distributed representation and neural networks continues to improve the state-of-the-art performance of natural language inference. However, it remains an open question whether the models perform real reasoning to reach their conclusions or rely on spurious correlations. Adversarial attacks have proven to be an important tool to help evaluate the Achilles{'} heel of the victim models. In this study, we explore the fundamental problem of developing attack models based on logic formalism. We propose NatLogAttack to perform systematic attacks centring around natural logic, a classical logic formalism that is traceable back to Aristotle{'}s syllogism and has been closely developed for natural language inference. The proposed framework renders both label-preserving and label-flipping attacks. We show that compared to the existing attack models, NatLogAttack generates better adversarial examples with fewer visits to the victim models. The victim models are found to be more vulnerable under the label-flipping setting. NatLogAttack provides a tool to probe the existing and future NLI models{'} capacity from a key viewpoint and we hope more logic-based attacks will be further explored for understanding the desired property of reasoning.",
}
| Reasoning has been a central topic in artificial intelligence from the beginning. The recent progress made on distributed representation and neural networks continues to improve the state-of-the-art performance of natural language inference. However, it remains an open question whether the models perform real reasoning to reach their conclusions or rely on spurious correlations. Adversarial attacks have proven to be an important tool to help evaluate the Achilles{'} heel of the victim models. In this study, we explore the fundamental problem of developing attack models based on logic formalism. We propose NatLogAttack to perform systematic attacks centring around natural logic, a classical logic formalism that is traceable back to Aristotle{'}s syllogism and has been closely developed for natural language inference. The proposed framework renders both label-preserving and label-flipping attacks. We show that compared to the existing attack models, NatLogAttack generates better adversarial examples with fewer visits to the victim models. The victim models are found to be more vulnerable under the label-flipping setting. NatLogAttack provides a tool to probe the existing and future NLI models{'} capacity from a key viewpoint and we hope more logic-based attacks will be further explored for understanding the desired property of reasoning. | [
"Zheng, Zi{'}ou",
"Zhu, Xiaodan"
] | NatLogAttack: A Framework for Attacking Natural Language Inference Models with Natural Logic | acl-long.554 | Poster | 2307.02849 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.555.bib | https://aclanthology.org/2023.acl-long.555/ | @inproceedings{sharma-etal-2023-cognitive,
title = "Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction",
author = "Sharma, Ashish and
Rushton, Kevin and
Lin, Inna and
Wadden, David and
Lucas, Khendra and
Miner, Adam and
Nguyen, Theresa and
Althoff, Tim",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.555",
doi = "10.18653/v1/2023.acl-long.555",
pages = "9977--10000",
abstract = "A proven therapeutic technique to overcome negative thoughts is to replace them with a more hopeful {``}reframed thought.{''} Although therapy can help people practice and learn this Cognitive Reframing of Negative Thoughts, clinician shortages and mental health stigma commonly limit people{'}s access to therapy. In this paper, we conduct a human-centered study of how language models may assist people in reframing negative thoughts. Based on psychology literature, we define a framework of seven linguistic attributes that can be used to reframe a thought. We develop automated metrics to measure these attributes and validate them with expert judgements from mental health practitioners. We collect a dataset of 600 situations, thoughts and reframes from practitioners and use it to train a retrieval-enhanced in-context learning model that effectively generates reframed thoughts and controls their linguistic attributes. To investigate what constitutes a {``}high-quality{''} reframe, we conduct an IRB-approved randomized field study on a large mental health website with over 2,000 participants. Amongst other findings, we show that people prefer highly empathic or specific reframes, as opposed to reframes that are overly positive. Our findings provide key implications for the use of LMs to assist people in overcoming negative thoughts.",
}
| A proven therapeutic technique to overcome negative thoughts is to replace them with a more hopeful {``}reframed thought.{''} Although therapy can help people practice and learn this Cognitive Reframing of Negative Thoughts, clinician shortages and mental health stigma commonly limit people{'}s access to therapy. In this paper, we conduct a human-centered study of how language models may assist people in reframing negative thoughts. Based on psychology literature, we define a framework of seven linguistic attributes that can be used to reframe a thought. We develop automated metrics to measure these attributes and validate them with expert judgements from mental health practitioners. We collect a dataset of 600 situations, thoughts and reframes from practitioners and use it to train a retrieval-enhanced in-context learning model that effectively generates reframed thoughts and controls their linguistic attributes. To investigate what constitutes a {``}high-quality{''} reframe, we conduct an IRB-approved randomized field study on a large mental health website with over 2,000 participants. Amongst other findings, we show that people prefer highly empathic or specific reframes, as opposed to reframes that are overly positive. Our findings provide key implications for the use of LMs to assist people in overcoming negative thoughts. | [
"Sharma, Ashish",
"Rushton, Kevin",
"Lin, Inna",
"Wadden, David",
"Lucas, Khendra",
"Miner, Adam",
"Nguyen, Theresa",
"Althoff, Tim"
] | Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction | acl-long.555 | Oral | 2305.02466 | [
"https://github.com/behavioral-data/cognitive-reframing-of-negative-thoughts"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.556.bib | https://aclanthology.org/2023.acl-long.556/ | @inproceedings{pavlopoulos-etal-2023-dating,
title = "Dating {G}reek Papyri with Text Regression",
author = "Pavlopoulos, John and
Konstantinidou, Maria and
Marthot-Santaniello, Isabelle and
Essler, Holger and
Paparigopoulou, Asimina",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.556",
doi = "10.18653/v1/2023.acl-long.556",
pages = "10001--10013",
abstract = "Dating Greek papyri accurately is crucial not only to edit their texts but also to understand numerous other aspects of ancient writing, document and book production and circulation, as well as various other aspects of administration, everyday life and intellectual history of antiquity. Although a substantial number of Greek papyri documents bear a date or other conclusive data as to their chronological placement, an even larger number can only be dated tentatively or in approximation, due to the lack of decisive evidence. By creating a dataset of 389 transcriptions of documentary Greek papyri, we train 389 regression models and we predict a date for the papyri with an average MAE of 54 years and an MSE of 1.17, outperforming image classifiers and other baselines. Last, we release date estimations for 159 manuscripts, for which only the upper limit is known.",
}
| Dating Greek papyri accurately is crucial not only to edit their texts but also to understand numerous other aspects of ancient writing, document and book production and circulation, as well as various other aspects of administration, everyday life and intellectual history of antiquity. Although a substantial number of Greek papyri documents bear a date or other conclusive data as to their chronological placement, an even larger number can only be dated tentatively or in approximation, due to the lack of decisive evidence. By creating a dataset of 389 transcriptions of documentary Greek papyri, we train 389 regression models and we predict a date for the papyri with an average MAE of 54 years and an MSE of 1.17, outperforming image classifiers and other baselines. Last, we release date estimations for 159 manuscripts, for which only the upper limit is known. | [
"Pavlopoulos, John",
"Konstantinidou, Maria",
"Marthot-Santaniello, Isabelle",
"Essler, Holger",
"Paparigopoulou, Asimina"
] | Dating Greek Papyri with Text Regression | acl-long.556 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.557.bib | https://aclanthology.org/2023.acl-long.557/ | @inproceedings{trivedi-etal-2023-interleaving,
title = "Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions",
author = "Trivedi, Harsh and
Balasubramanian, Niranjan and
Khot, Tushar and
Sabharwal, Ashish",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.557",
doi = "10.18653/v1/2023.acl-long.557",
pages = "10014--10037",
abstract = "Prompting-based large language models (LLMs) are surprisingly powerful at generating natural language reasoning steps or Chains-of-Thoughts (CoT) for multi-step question answering (QA). They struggle, however, when the necessary knowledge is either unavailable to the LLM or not up-to-date within its parameters. While using the question to retrieve relevant text from an external knowledge source helps LLMs, we observe that this one-step retrieve-and-read approach is insufficient for multi-step QA. Here, \textit{what to retrieve} depends on \textit{what has already been derived}, which in turn may depend on \textit{what was previously retrieved}. To address this, we propose IRCoT, a new approach for multi-step QA that interleaves retrieval with steps (sentences) in a CoT, guiding the retrieval with CoT and in turn using retrieved results to improve CoT. Using IRCoT with GPT3 substantially improves retrieval (up to 21 points) as well as downstream QA (up to 15 points) on four datasets: HotpotQA, 2WikiMultihopQA, MuSiQue, and IIRC. We observe similar substantial gains in out-of-distribution (OOD) settings as well as with much smaller models such as Flan-T5-large without additional training. IRCoT reduces model hallucination, resulting in factually more accurate CoT reasoning.",
}
| Prompting-based large language models (LLMs) are surprisingly powerful at generating natural language reasoning steps or Chains-of-Thoughts (CoT) for multi-step question answering (QA). They struggle, however, when the necessary knowledge is either unavailable to the LLM or not up-to-date within its parameters. While using the question to retrieve relevant text from an external knowledge source helps LLMs, we observe that this one-step retrieve-and-read approach is insufficient for multi-step QA. Here, \textit{what to retrieve} depends on \textit{what has already been derived}, which in turn may depend on \textit{what was previously retrieved}. To address this, we propose IRCoT, a new approach for multi-step QA that interleaves retrieval with steps (sentences) in a CoT, guiding the retrieval with CoT and in turn using retrieved results to improve CoT. Using IRCoT with GPT3 substantially improves retrieval (up to 21 points) as well as downstream QA (up to 15 points) on four datasets: HotpotQA, 2WikiMultihopQA, MuSiQue, and IIRC. We observe similar substantial gains in out-of-distribution (OOD) settings as well as with much smaller models such as Flan-T5-large without additional training. IRCoT reduces model hallucination, resulting in factually more accurate CoT reasoning. | [
"Trivedi, Harsh",
"Balasubramanian, Niranjan",
"Khot, Tushar",
"Sabharwal, Ashish"
] | Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions | acl-long.557 | Poster | 2212.10509 | [
"https://github.com/stonybrooknlp/ircot"
] | https://huggingface.co/papers/2212.10509 | 0 | 0 | 0 | 4 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.558.bib | https://aclanthology.org/2023.acl-long.558/ | @inproceedings{baek-etal-2023-direct,
title = "Direct Fact Retrieval from Knowledge Graphs without Entity Linking",
author = "Baek, Jinheon and
Aji, Alham Fikri and
Lehmann, Jens and
Hwang, Sung Ju",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.558",
doi = "10.18653/v1/2023.acl-long.558",
pages = "10038--10055",
abstract = "There has been a surge of interest in utilizing Knowledge Graphs (KGs) for various natural language processing/understanding tasks. The conventional mechanism to retrieve facts in KGs usually involves three steps: entity span detection, entity disambiguation, and relation classification. However, this approach requires additional labels for training each of the three subcomponents in addition to pairs of input texts and facts, and also may accumulate errors propagated from failures in previous steps. To tackle these limitations, we propose a simple knowledge retrieval framework, which directly retrieves facts from the KGs given the input text based on their representational similarities, which we refer to as Direct Fact Retrieval (DiFaR). Specifically, we first embed all facts in KGs onto a dense embedding space by using a language model trained by only pairs of input texts and facts, and then provide the nearest facts in response to the input text. Since the fact, consisting of only two entities and one relation, has little context to encode, we propose to further refine ranks of top-k retrieved facts with a reranker that contextualizes the input text and the fact jointly. We validate our DiFaR framework on multiple fact retrieval tasks, showing that it significantly outperforms relevant baselines that use the three-step approach.",
}
| There has been a surge of interest in utilizing Knowledge Graphs (KGs) for various natural language processing/understanding tasks. The conventional mechanism to retrieve facts in KGs usually involves three steps: entity span detection, entity disambiguation, and relation classification. However, this approach requires additional labels for training each of the three subcomponents in addition to pairs of input texts and facts, and also may accumulate errors propagated from failures in previous steps. To tackle these limitations, we propose a simple knowledge retrieval framework, which directly retrieves facts from the KGs given the input text based on their representational similarities, which we refer to as Direct Fact Retrieval (DiFaR). Specifically, we first embed all facts in KGs onto a dense embedding space by using a language model trained by only pairs of input texts and facts, and then provide the nearest facts in response to the input text. Since the fact, consisting of only two entities and one relation, has little context to encode, we propose to further refine ranks of top-k retrieved facts with a reranker that contextualizes the input text and the fact jointly. We validate our DiFaR framework on multiple fact retrieval tasks, showing that it significantly outperforms relevant baselines that use the three-step approach. | [
"Baek, Jinheon",
"Aji, Alham Fikri",
"Lehmann, Jens",
"Hwang, Sung Ju"
] | Direct Fact Retrieval from Knowledge Graphs without Entity Linking | acl-long.558 | Poster | 2305.12416 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.559.bib | https://aclanthology.org/2023.acl-long.559/ | @inproceedings{neeman-etal-2023-disentqa,
title = "{D}isent{QA}: Disentangling Parametric and Contextual Knowledge with Counterfactual Question Answering",
author = "Neeman, Ella and
Aharoni, Roee and
Honovich, Or and
Choshen, Leshem and
Szpektor, Idan and
Abend, Omri",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.559",
doi = "10.18653/v1/2023.acl-long.559",
pages = "10056--10070",
abstract = "Question answering models commonly have access to two sources of {``}knowledge{''} during inference time: (1) parametric knowledge - the factual knowledge encoded in the model weights, and (2) contextual knowledge - external knowledge (e.g., a Wikipedia passage) given to the model to generate a grounded answer. Having these two sources of knowledge entangled together is a core issue for generative QA models as it is unclear whether the answer stems from the given non-parametric knowledge or not. This unclarity has implications on issues of trust, interpretability and factuality. In this work, we propose a new paradigm in which QA models are trained to disentangle the two sources of knowledge. Using counterfactual data augmentation, we introduce a model that predicts two answers for a given question: one based on given contextual knowledge and one based on parametric knowledge. Our experiments on the Natural Questions dataset show that this approach improves the performance of QA models by making them more robust to knowledge conflicts between the two knowledge sources, while generating useful disentangled answers.",
}
| Question answering models commonly have access to two sources of {``}knowledge{''} during inference time: (1) parametric knowledge - the factual knowledge encoded in the model weights, and (2) contextual knowledge - external knowledge (e.g., a Wikipedia passage) given to the model to generate a grounded answer. Having these two sources of knowledge entangled together is a core issue for generative QA models as it is unclear whether the answer stems from the given non-parametric knowledge or not. This unclarity has implications on issues of trust, interpretability and factuality. In this work, we propose a new paradigm in which QA models are trained to disentangle the two sources of knowledge. Using counterfactual data augmentation, we introduce a model that predicts two answers for a given question: one based on given contextual knowledge and one based on parametric knowledge. Our experiments on the Natural Questions dataset show that this approach improves the performance of QA models by making them more robust to knowledge conflicts between the two knowledge sources, while generating useful disentangled answers. | [
"Neeman, Ella",
"Aharoni, Roee",
"Honovich, Or",
"Choshen, Leshem",
"Szpektor, Idan",
"Abend, Omri"
] | DisentQA: Disentangling Parametric and Contextual Knowledge with Counterfactual Question Answering | acl-long.559 | Poster | 2211.05655 | [
"https://github.com/ellaneeman/disent_qa"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.560.bib | https://aclanthology.org/2023.acl-long.560/ | @inproceedings{li-etal-2023-new,
title = "A New Direction in Stance Detection: Target-Stance Extraction in the Wild",
author = "Li, Yingjie and
Garg, Krishna and
Caragea, Cornelia",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.560",
doi = "10.18653/v1/2023.acl-long.560",
pages = "10071--10085",
abstract = "Stance detection aims to detect the stance toward a corresponding target. Existing works use the assumption that the target is known in advance, which is often not the case in the wild. Given a text from social media platforms, the target information is often unknown due to implicit mentions in the source text and it is infeasible to have manual target annotations at a large scale. Therefore, in this paper, we propose a new task Target-Stance Extraction (TSE) that aims to extract the (target, stance) pair from the text. We benchmark the task by proposing a two-stage framework that first identifies the relevant target in the text and then detects the stance given the predicted target and text. Specifically, we first propose two different settings: Target Classification and Target Generation, to identify the potential target from a given text. Then we propose a multi-task approach that takes target prediction as the auxiliary task to detect the stance toward the predicted target. We evaluate the proposed framework on both in-target stance detection in which the test target is always seen in the training stage and zero-shot stance detection that needs to detect the stance for the targets that are unseen during the training phase. The new TSE task can facilitate future research in the field of stance detection.",
}
| Stance detection aims to detect the stance toward a corresponding target. Existing works use the assumption that the target is known in advance, which is often not the case in the wild. Given a text from social media platforms, the target information is often unknown due to implicit mentions in the source text and it is infeasible to have manual target annotations at a large scale. Therefore, in this paper, we propose a new task Target-Stance Extraction (TSE) that aims to extract the (target, stance) pair from the text. We benchmark the task by proposing a two-stage framework that first identifies the relevant target in the text and then detects the stance given the predicted target and text. Specifically, we first propose two different settings: Target Classification and Target Generation, to identify the potential target from a given text. Then we propose a multi-task approach that takes target prediction as the auxiliary task to detect the stance toward the predicted target. We evaluate the proposed framework on both in-target stance detection in which the test target is always seen in the training stage and zero-shot stance detection that needs to detect the stance for the targets that are unseen during the training phase. The new TSE task can facilitate future research in the field of stance detection. | [
"Li, Yingjie",
"Garg, Krishna",
"Caragea, Cornelia"
] | A New Direction in Stance Detection: Target-Stance Extraction in the Wild | acl-long.560 | Oral | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.561.bib | https://aclanthology.org/2023.acl-long.561/ | @inproceedings{le-etal-2023-improved,
title = "Improved Instruction Ordering in Recipe-Grounded Conversation",
author = "Le, Duong and
Guo, Ruohao and
Xu, Wei and
Ritter, Alan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.561",
doi = "10.18653/v1/2023.acl-long.561",
pages = "10086--10104",
abstract = "In this paper, we study the task of instructional dialogue and focus on the cooking domain. Analyzing the generated output of the GPT-J model, we reveal that the primary challenge for a recipe-grounded dialog system is how to provide the instructions in the correct order. We hypothesize that this is due to the model{'}s lack of understanding of user intent and inability to track the instruction state (i.e., which step was last instructed). Therefore, we propose to explore two auxiliary subtasks, namely User Intent Detection and Instruction State Tracking, to support Response Generation with improved instruction grounding. Experimenting with our newly collected dataset, ChattyChef, shows that incorporating user intent and instruction state information helps the response generation model mitigate the incorrect order issue. Furthermore, to investigate whether ChatGPT has completely solved this task, we analyze its outputs and find that it also makes mistakes (10.7{\%} of the responses), about half of which are out-of-order instructions. We will release ChattyChef to facilitate further research in this area at: \url{https://github.com/octaviaguo/ChattyChef}.",
}
| In this paper, we study the task of instructional dialogue and focus on the cooking domain. Analyzing the generated output of the GPT-J model, we reveal that the primary challenge for a recipe-grounded dialog system is how to provide the instructions in the correct order. We hypothesize that this is due to the model{'}s lack of understanding of user intent and inability to track the instruction state (i.e., which step was last instructed). Therefore, we propose to explore two auxiliary subtasks, namely User Intent Detection and Instruction State Tracking, to support Response Generation with improved instruction grounding. Experimenting with our newly collected dataset, ChattyChef, shows that incorporating user intent and instruction state information helps the response generation model mitigate the incorrect order issue. Furthermore, to investigate whether ChatGPT has completely solved this task, we analyze its outputs and find that it also makes mistakes (10.7{\%} of the responses), about half of which are out-of-order instructions. We will release ChattyChef to facilitate further research in this area at: \url{https://github.com/octaviaguo/ChattyChef}. | [
"Le, Duong",
"Guo, Ruohao",
"Xu, Wei",
"Ritter, Alan"
] | Improved Instruction Ordering in Recipe-Grounded Conversation | acl-long.561 | Poster | 2305.17280 | [
"https://github.com/octaviaguo/chattychef"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.562.bib | https://aclanthology.org/2023.acl-long.562/ | @inproceedings{oh-schuler-2023-token,
title = "Token-wise Decomposition of Autoregressive Language Model Hidden States for Analyzing Model Predictions",
author = "Oh, Byung-Doh and
Schuler, William",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.562",
doi = "10.18653/v1/2023.acl-long.562",
pages = "10105--10117",
abstract = "While there is much recent interest in studying why Transformer-based large language models make predictions the way they do, the complex computations performed within each layer have made their behavior somewhat opaque. To mitigate this opacity, this work presents a linear decomposition of final hidden states from autoregressive language models based on each initial input token, which is exact for virtually all contemporary Transformer architectures. This decomposition allows the definition of probability distributions that ablate the contribution of specific input tokens, which can be used to analyze their influence on model probabilities over a sequence of upcoming words with only one forward pass from the model. Using the change in next-word probability as a measure of importance, this work first examines which context words make the biggest contribution to language model predictions. Regression experiments suggest that Transformer-based language models rely primarily on collocational associations, followed by linguistic factors such as syntactic dependencies and coreference relationships in making next-word predictions. Additionally, analyses using these measures to predict syntactic dependencies and coreferent mention spans show that collocational association and repetitions of the same token largely explain the language models{'} predictions on these tasks.",
}
| While there is much recent interest in studying why Transformer-based large language models make predictions the way they do, the complex computations performed within each layer have made their behavior somewhat opaque. To mitigate this opacity, this work presents a linear decomposition of final hidden states from autoregressive language models based on each initial input token, which is exact for virtually all contemporary Transformer architectures. This decomposition allows the definition of probability distributions that ablate the contribution of specific input tokens, which can be used to analyze their influence on model probabilities over a sequence of upcoming words with only one forward pass from the model. Using the change in next-word probability as a measure of importance, this work first examines which context words make the biggest contribution to language model predictions. Regression experiments suggest that Transformer-based language models rely primarily on collocational associations, followed by linguistic factors such as syntactic dependencies and coreference relationships in making next-word predictions. Additionally, analyses using these measures to predict syntactic dependencies and coreferent mention spans show that collocational association and repetitions of the same token largely explain the language models{'} predictions on these tasks. | [
"Oh, Byung-Doh",
"Schuler, William"
] | Token-wise Decomposition of Autoregressive Language Model Hidden States for Analyzing Model Predictions | acl-long.562 | Poster | 2305.10614 | [
"https://github.com/byungdoh/llm_decomposition"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.563.bib | https://aclanthology.org/2023.acl-long.563/ | @inproceedings{wang-etal-2023-document,
title = "Document-Level Multi-Event Extraction with Event Proxy Nodes and Hausdorff Distance Minimization",
author = "Wang, Xinyu and
Gui, Lin and
He, Yulan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.563",
doi = "10.18653/v1/2023.acl-long.563",
pages = "10118--10133",
abstract = "Document-level multi-event extraction aims to extract the structural information from a given document automatically. Most recent approaches usually involve two steps: (1) modeling entity interactions; (2) decoding entity interactions into events. However, such approaches ignore a global view of inter-dependency of multiple events. Moreover, an event is decoded by iteratively merging its related entities as arguments, which might suffer from error propagation and is computationally inefficient. In this paper, we propose an alternative approach for document-level multi-event extraction with event proxy nodes and Hausdorff distance minimization. The event proxy nodes, representing pseudo-events, are able to build connections with other event proxy nodes, essentially capturing global information. The Hausdorff distance makes it possible to compare the similarity between the set of predicted events and the set of ground-truth events. By directly minimizing Hausdorff distance, the model is trained towards the global optimum directly, which improves performance and reduces training time. Experimental results show that our model outperforms previous state-of-the-art method in F1-score on two datasets with only a fraction of training time.",
}
| Document-level multi-event extraction aims to extract the structural information from a given document automatically. Most recent approaches usually involve two steps: (1) modeling entity interactions; (2) decoding entity interactions into events. However, such approaches ignore a global view of inter-dependency of multiple events. Moreover, an event is decoded by iteratively merging its related entities as arguments, which might suffer from error propagation and is computationally inefficient. In this paper, we propose an alternative approach for document-level multi-event extraction with event proxy nodes and Hausdorff distance minimization. The event proxy nodes, representing pseudo-events, are able to build connections with other event proxy nodes, essentially capturing global information. The Hausdorff distance makes it possible to compare the similarity between the set of predicted events and the set of ground-truth events. By directly minimizing Hausdorff distance, the model is trained towards the global optimum directly, which improves performance and reduces training time. Experimental results show that our model outperforms previous state-of-the-art method in F1-score on two datasets with only a fraction of training time. | [
"Wang, Xinyu",
"Gui, Lin",
"He, Yulan"
] | Document-Level Multi-Event Extraction with Event Proxy Nodes and Hausdorff Distance Minimization | acl-long.563 | Poster | 2305.18926 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.564.bib | https://aclanthology.org/2023.acl-long.564/ | @inproceedings{zhang-etal-2023-dialog,
title = "Dialog-Post: Multi-Level Self-Supervised Objectives and Hierarchical Model for Dialogue Post-Training",
author = "Zhang, Zhenyu and
Shen, Lei and
Zhao, Yuming and
Chen, Meng and
He, Xiaodong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.564",
doi = "10.18653/v1/2023.acl-long.564",
pages = "10134--10148",
abstract = "Dialogue representation and understanding aim to convert conversational inputs into embeddings and fulfill discriminative tasks. Compared with free-form text, dialogue has two important characteristics, hierarchical semantic structure and multi-facet attributes. Therefore, directly applying the pretrained language models (PLMs) might result in unsatisfactory performance. Recently, several work focused on the dialogue-adaptive post-training (DialPost) that further trains PLMs to fit dialogues. To model dialogues more comprehensively, we propose a DialPost method, Dialog-Post, with multi-level self-supervised objectives and a hierarchical model. These objectives leverage dialogue-specific attributes and use self-supervised signals to fully facilitate the representation and understanding of dialogues. The novel model is a hierarchical segment-wise self-attention network, which contains inner-segment and inter-segment self-attention sub-layers followed by an aggregation and updating module. To evaluate the effectiveness of our methods, we first apply two public datasets for the verification of representation ability. Then we conduct experiments on a newly-labelled dataset that is annotated with 4 dialogue understanding tasks. Experimental results show that our method outperforms existing SOTA models and achieves a 3.3{\%} improvement on average.",
}
| Dialogue representation and understanding aim to convert conversational inputs into embeddings and fulfill discriminative tasks. Compared with free-form text, dialogue has two important characteristics, hierarchical semantic structure and multi-facet attributes. Therefore, directly applying the pretrained language models (PLMs) might result in unsatisfactory performance. Recently, several work focused on the dialogue-adaptive post-training (DialPost) that further trains PLMs to fit dialogues. To model dialogues more comprehensively, we propose a DialPost method, Dialog-Post, with multi-level self-supervised objectives and a hierarchical model. These objectives leverage dialogue-specific attributes and use self-supervised signals to fully facilitate the representation and understanding of dialogues. The novel model is a hierarchical segment-wise self-attention network, which contains inner-segment and inter-segment self-attention sub-layers followed by an aggregation and updating module. To evaluate the effectiveness of our methods, we first apply two public datasets for the verification of representation ability. Then we conduct experiments on a newly-labelled dataset that is annotated with 4 dialogue understanding tasks. Experimental results show that our method outperforms existing SOTA models and achieves a 3.3{\%} improvement on average. | [
"Zhang, Zhenyu",
"Shen, Lei",
"Zhao, Yuming",
"Chen, Meng",
"He, Xiaodong"
] | Dialog-Post: Multi-Level Self-Supervised Objectives and Hierarchical Model for Dialogue Post-Training | acl-long.564 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.565.bib | https://aclanthology.org/2023.acl-long.565/ | @inproceedings{kwak-etal-2023-language,
title = "Language Detoxification with Attribute-Discriminative Latent Space",
author = "Kwak, Jin Myung and
Kim, Minseon and
Hwang, Sung Ju",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.565",
doi = "10.18653/v1/2023.acl-long.565",
pages = "10149--10171",
abstract = "Transformer-based Language Models (LMs) have achieved impressive results on natural language understanding tasks, but they can also generate toxic text such as insults, threats, and profanity, limiting their real-world applications. To overcome this issue, a few text generation approaches aim to detoxify toxic texts using additional LMs or perturbations. However, previous methods require excessive memory, computations, and time which are serious bottlenecks in their real-world application. To address such limitations, we propose an effective yet efficient method for language detoxification using an attribute-discriminative latent space. Specifically, we project the latent space of an original Transformer LM onto a discriminative latent space that well-separates texts by their attributes using a projection block and an attribute discriminator. This allows the LM to control the text generation to be non-toxic with minimal memory and computation overhead. We validate our model, Attribute-Discriminative Language Model (ADLM) on detoxified language and dialogue generation tasks, on which our method significantly outperforms baselines both in performance and efficiency.",
}
| Transformer-based Language Models (LMs) have achieved impressive results on natural language understanding tasks, but they can also generate toxic text such as insults, threats, and profanity, limiting their real-world applications. To overcome this issue, a few text generation approaches aim to detoxify toxic texts using additional LMs or perturbations. However, previous methods require excessive memory, computations, and time which are serious bottlenecks in their real-world application. To address such limitations, we propose an effective yet efficient method for language detoxification using an attribute-discriminative latent space. Specifically, we project the latent space of an original Transformer LM onto a discriminative latent space that well-separates texts by their attributes using a projection block and an attribute discriminator. This allows the LM to control the text generation to be non-toxic with minimal memory and computation overhead. We validate our model, Attribute-Discriminative Language Model (ADLM) on detoxified language and dialogue generation tasks, on which our method significantly outperforms baselines both in performance and efficiency. | [
"Kwak, Jin Myung",
"Kim, Minseon",
"Hwang, Sung Ju"
] | Language Detoxification with Attribute-Discriminative Latent Space | acl-long.565 | Poster | 2210.10329 | [
"https://github.com/jin8/adlm"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.566.bib | https://aclanthology.org/2023.acl-long.566/ | @inproceedings{min-etal-2023-just,
title = "Just Like a Human Would, Direct Access to Sarcasm Augmented with Potential Result and Reaction",
author = "Min, Changrong and
Li, Ximing and
Yang, Liang and
Wang, Zhilin and
Xu, Bo and
Lin, Hongfei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.566",
doi = "10.18653/v1/2023.acl-long.566",
pages = "10172--10183",
abstract = "Sarcasm, as a form of irony conveying mockery and contempt, has been widespread in social media such as Twitter and Weibo, where the sarcastic text is commonly characterized as an incongruity between the surface positive and negative situation. Naturally, it has an urgent demand to automatically identify sarcasm from social media, so as to illustrate people{'}s real views toward specific targets. In this paper, we develop a novel sarcasm detection method, namely Sarcasm Detector with Augmentation of Potential Result and Reaction (SD-APRR). Inspired by the direct access view, we treat each sarcastic text as an incomplete version without latent content associated with implied negative situations, including the result and human reaction caused by its observable content. To fill the latent content, we estimate the potential result and human reaction for each given training sample by [xEffect] and [xReact] relations inferred by the pre-trained commonsense reasoning tool COMET, and integrate the sample with them as an augmented one. We can then employ those augmented samples to train the sarcasm detector, whose encoder is a graph neural network with a denoising module. We conduct extensive empirical experiments to evaluate the effectiveness of SD-APRR. The results demonstrate that SD-APRR can outperform strong baselines on benchmark datasets.",
}
| Sarcasm, as a form of irony conveying mockery and contempt, has been widespread in social media such as Twitter and Weibo, where the sarcastic text is commonly characterized as an incongruity between the surface positive and negative situation. Naturally, it has an urgent demand to automatically identify sarcasm from social media, so as to illustrate people{'}s real views toward specific targets. In this paper, we develop a novel sarcasm detection method, namely Sarcasm Detector with Augmentation of Potential Result and Reaction (SD-APRR). Inspired by the direct access view, we treat each sarcastic text as an incomplete version without latent content associated with implied negative situations, including the result and human reaction caused by its observable content. To fill the latent content, we estimate the potential result and human reaction for each given training sample by [xEffect] and [xReact] relations inferred by the pre-trained commonsense reasoning tool COMET, and integrate the sample with them as an augmented one. We can then employ those augmented samples to train the sarcasm detector, whose encoder is a graph neural network with a denoising module. We conduct extensive empirical experiments to evaluate the effectiveness of SD-APRR. The results demonstrate that SD-APRR can outperform strong baselines on benchmark datasets. | [
"Min, Changrong",
"Li, Ximing",
"Yang, Liang",
"Wang, Zhilin",
"Xu, Bo",
"Lin, Hongfei"
] | Just Like a Human Would, Direct Access to Sarcasm Augmented with Potential Result and Reaction | acl-long.566 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.567.bib | https://aclanthology.org/2023.acl-long.567/ | @inproceedings{cui-sachan-2023-adaptive,
title = "Adaptive and Personalized Exercise Generation for Online Language Learning",
author = "Cui, Peng and
Sachan, Mrinmaya",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.567",
doi = "10.18653/v1/2023.acl-long.567",
pages = "10184--10198",
abstract = "Adaptive learning aims to provide customized educational activities (e.g., exercises) to address individual learning needs. However, manual construction and delivery of such activities is a laborious process. Thus, in this paper, we study a novel task of adaptive and personalized exercise generation for online language learning. To this end, we combine a knowledge tracing model that estimates each student{'}s evolving knowledge states from their learning history and a controlled text generation model that generates exercise sentences based on the student{'}s current estimated knowledge state and instructor requirements of desired properties (e.g., domain knowledge and difficulty). We train and evaluate our model on real-world learner interaction data from Duolingo and demonstrate that LMs guided by student states can generate superior exercises. Then, we discuss the potential use of our model in educational applications using various simulations. These simulations show that our model can adapt to students{'} individual abilities and can facilitate their learning efficiency by personalizing learning sequences.",
}
| Adaptive learning aims to provide customized educational activities (e.g., exercises) to address individual learning needs. However, manual construction and delivery of such activities is a laborious process. Thus, in this paper, we study a novel task of adaptive and personalized exercise generation for online language learning. To this end, we combine a knowledge tracing model that estimates each student{'}s evolving knowledge states from their learning history and a controlled text generation model that generates exercise sentences based on the student{'}s current estimated knowledge state and instructor requirements of desired properties (e.g., domain knowledge and difficulty). We train and evaluate our model on real-world learner interaction data from Duolingo and demonstrate that LMs guided by student states can generate superior exercises. Then, we discuss the potential use of our model in educational applications using various simulations. These simulations show that our model can adapt to students{'} individual abilities and can facilitate their learning efficiency by personalizing learning sequences. | [
"Cui, Peng",
"Sachan, Mrinmaya"
] | Adaptive and Personalized Exercise Generation for Online Language Learning | acl-long.567 | Oral | 2306.02457 | [
"https://github.com/nlpcui/adaptiveqg"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.568.bib | https://aclanthology.org/2023.acl-long.568/ | @inproceedings{storks-etal-2023-nlp,
title = "{NLP} Reproducibility For All: Understanding Experiences of Beginners",
author = "Storks, Shane and
Yu, Keunwoo and
Ma, Ziqiao and
Chai, Joyce",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.568",
doi = "10.18653/v1/2023.acl-long.568",
pages = "10199--10219",
abstract = "As natural language processing (NLP) has recently seen an unprecedented level of excitement, and more people are eager to enter the field, it is unclear whether current research reproducibility efforts are sufficient for this group of beginners to apply the latest developments. To understand their needs, we conducted a study with 93 students in an introductory NLP course, where students reproduced the results of recent NLP papers. Surprisingly, we find that their programming skill and comprehension of research papers have a limited impact on their effort spent completing the exercise. Instead, we find accessibility efforts by research authors to be the key to success, including complete documentation, better coding practice, and easier access to data files. Going forward, we recommend that NLP researchers pay close attention to these simple aspects of open-sourcing their work, and use insights from beginners{'} feedback to provide actionable ideas on how to better support them.",
}
| As natural language processing (NLP) has recently seen an unprecedented level of excitement, and more people are eager to enter the field, it is unclear whether current research reproducibility efforts are sufficient for this group of beginners to apply the latest developments. To understand their needs, we conducted a study with 93 students in an introductory NLP course, where students reproduced the results of recent NLP papers. Surprisingly, we find that their programming skill and comprehension of research papers have a limited impact on their effort spent completing the exercise. Instead, we find accessibility efforts by research authors to be the key to success, including complete documentation, better coding practice, and easier access to data files. Going forward, we recommend that NLP researchers pay close attention to these simple aspects of open-sourcing their work, and use insights from beginners{'} feedback to provide actionable ideas on how to better support them. | [
"Storks, Shane",
"Yu, Keunwoo",
"Ma, Ziqiao",
"Chai, Joyce"
] | NLP Reproducibility For All: Understanding Experiences of Beginners | acl-long.568 | Poster | 2305.16579 | [
"https://github.com/sled-group/nlp-reproducibility-for-beginners"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.569.bib | https://aclanthology.org/2023.acl-long.569/ | @inproceedings{stengel-eskin-etal-2023-chicken,
title = "Why Did the Chicken Cross the Road? Rephrasing and Analyzing Ambiguous Questions in {VQA}",
author = "Stengel-Eskin, Elias and
Guallar-Blasco, Jimena and
Zhou, Yi and
Van Durme, Benjamin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.569",
doi = "10.18653/v1/2023.acl-long.569",
pages = "10220--10237",
abstract = "Natural language is ambiguous. Resolving ambiguous questions is key to successfully answering them. Focusing on questions about images, we create a dataset of ambiguous examples. We annotate these, grouping answers by the underlying question they address and rephrasing the question for each group to reduce ambiguity. Our analysis reveals a linguistically-aligned ontology of reasons for ambiguity in visual questions. We then develop an English question-generation model which we demonstrate via automatic and human evaluation produces less ambiguous questions. We further show that the question generation objective we use allows the model to integrate answer group information without any direct supervision.",
}
| Natural language is ambiguous. Resolving ambiguous questions is key to successfully answering them. Focusing on questions about images, we create a dataset of ambiguous examples. We annotate these, grouping answers by the underlying question they address and rephrasing the question for each group to reduce ambiguity. Our analysis reveals a linguistically-aligned ontology of reasons for ambiguity in visual questions. We then develop an English question-generation model which we demonstrate via automatic and human evaluation produces less ambiguous questions. We further show that the question generation objective we use allows the model to integrate answer group information without any direct supervision. | [
"Stengel-Eskin, Elias",
"Guallar-Blasco, Jimena",
"Zhou, Yi",
"Van Durme, Benjamin"
] | Why Did the Chicken Cross the Road? Rephrasing and Analyzing Ambiguous Questions in VQA | acl-long.569 | Oral | 2211.07516 | [
"https://github.com/esteng/ambiguous_vqa"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.570.bib | https://aclanthology.org/2023.acl-long.570/ | @inproceedings{he-etal-2023-umrspell,
title = "{UMRS}pell: Unifying the Detection and Correction Parts of Pre-trained Models towards {C}hinese Missing, Redundant, and Spelling Correction",
author = "He, Zheyu and
Zhu, Yujin and
Wang, Linlin and
Xu, Liang",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.570",
doi = "10.18653/v1/2023.acl-long.570",
pages = "10238--10250",
abstract = "Chinese Spelling Correction (CSC) is the task of detecting and correcting misspelled charac- ters in Chinese texts. As an important step for various downstream tasks, CSC confronts two challenges: 1) Character-level errors consist not only of spelling errors but also of missing and redundant ones that cause variable length between input and output texts, for which most CSC methods could not handle well because of the consistence length of texts required by their inherent detection-correction framework. Con- sequently, the two errors are considered out- side the scope and left to future work, despite the fact that they are widely found and bound to CSC task in Chinese industrial scenario, such as Automatic Speech Recognition (ASR) and Optical Character Recognition (OCR). 2) Most existing CSC methods focus on either detector or corrector and train different mod- els for each one, respectively, leading to in- sufficiency of parameters sharing. To address these issues, we propose a novel model UMR- Spell to learn detection and correction parts together at the same time from a multi-task learning perspective by using a detection trans- mission self-attention matrix, and flexibly deal with both missing, redundant, and spelling er- rors through re-tagging rules. Furthermore, we build a new dataset ECMR-2023 containing five kinds of character-level errors to enrich the CSC task closer to real-world applications. Ex- periments on both SIGHAN benchmarks and ECMR-2023 demonstrate the significant effec- tiveness of UMRSpell over previous represen- tative baselines.",
}
| Chinese Spelling Correction (CSC) is the task of detecting and correcting misspelled charac- ters in Chinese texts. As an important step for various downstream tasks, CSC confronts two challenges: 1) Character-level errors consist not only of spelling errors but also of missing and redundant ones that cause variable length between input and output texts, for which most CSC methods could not handle well because of the consistence length of texts required by their inherent detection-correction framework. Con- sequently, the two errors are considered out- side the scope and left to future work, despite the fact that they are widely found and bound to CSC task in Chinese industrial scenario, such as Automatic Speech Recognition (ASR) and Optical Character Recognition (OCR). 2) Most existing CSC methods focus on either detector or corrector and train different mod- els for each one, respectively, leading to in- sufficiency of parameters sharing. To address these issues, we propose a novel model UMR- Spell to learn detection and correction parts together at the same time from a multi-task learning perspective by using a detection trans- mission self-attention matrix, and flexibly deal with both missing, redundant, and spelling er- rors through re-tagging rules. Furthermore, we build a new dataset ECMR-2023 containing five kinds of character-level errors to enrich the CSC task closer to real-world applications. Ex- periments on both SIGHAN benchmarks and ECMR-2023 demonstrate the significant effec- tiveness of UMRSpell over previous represen- tative baselines. | [
"He, Zheyu",
"Zhu, Yujin",
"Wang, Linlin",
"Xu, Liang"
] | UMRSpell: Unifying the Detection and Correction Parts of Pre-trained Models towards Chinese Missing, Redundant, and Spelling Correction | acl-long.570 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.571.bib | https://aclanthology.org/2023.acl-long.571/ | @inproceedings{milbauer-etal-2023-lait,
title = "{LAIT}: Efficient Multi-Segment Encoding in Transformers with Layer-Adjustable Interaction",
author = "Milbauer, Jeremiah and
Louis, Annie and
Hosseini, Mohammad Javad and
Fabrikant, Alex and
Metzler, Donald and
Schuster, Tal",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.571",
doi = "10.18653/v1/2023.acl-long.571",
pages = "10251--10269",
abstract = "Transformer encoders contextualize token representations by attending to all other tokens at each layer, leading to quadratic increase in compute effort with the input length. In practice, however, the input text of many NLP tasks can be seen as a sequence of related segments (e.g., the sequence of sentences within a passage, or the hypothesis and premise in NLI). While attending across these segments is highly beneficial for many tasks, we hypothesize that this interaction can be delayed until later encoding stages. To this end, we introduce Layer-Adjustable Interactions in Transformers (LAIT). Within LAIT, segmented inputs are first encoded independently, and then jointly. This partial two-tower architecture bridges the gap between a Dual Encoder{'}s ability to pre-compute representations for segments and a fully self-attentive Transformer{'}s capacity to model cross-segment attention. The LAIT framework effectively leverages existing pretrained Transformers and converts them into the hybrid of the two aforementioned architectures, allowing for easy and intuitive control over the performance-efficiency tradeoff. Experimenting on a wide range of NLP tasks, we find LAIT able to reduce 30-50{\%} of the attention FLOPs on many tasks, while preserving high accuracy; in some practical settings, LAIT could reduce actual latency by orders of magnitude.",
}
| Transformer encoders contextualize token representations by attending to all other tokens at each layer, leading to quadratic increase in compute effort with the input length. In practice, however, the input text of many NLP tasks can be seen as a sequence of related segments (e.g., the sequence of sentences within a passage, or the hypothesis and premise in NLI). While attending across these segments is highly beneficial for many tasks, we hypothesize that this interaction can be delayed until later encoding stages. To this end, we introduce Layer-Adjustable Interactions in Transformers (LAIT). Within LAIT, segmented inputs are first encoded independently, and then jointly. This partial two-tower architecture bridges the gap between a Dual Encoder{'}s ability to pre-compute representations for segments and a fully self-attentive Transformer{'}s capacity to model cross-segment attention. The LAIT framework effectively leverages existing pretrained Transformers and converts them into the hybrid of the two aforementioned architectures, allowing for easy and intuitive control over the performance-efficiency tradeoff. Experimenting on a wide range of NLP tasks, we find LAIT able to reduce 30-50{\%} of the attention FLOPs on many tasks, while preserving high accuracy; in some practical settings, LAIT could reduce actual latency by orders of magnitude. | [
"Milbauer, Jeremiah",
"Louis, Annie",
"Hosseini, Mohammad Javad",
"Fabrikant, Alex",
"Metzler, Donald",
"Schuster, Tal"
] | LAIT: Efficient Multi-Segment Encoding in Transformers with Layer-Adjustable Interaction | acl-long.571 | Poster | 2305.19585 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.572.bib | https://aclanthology.org/2023.acl-long.572/ | @inproceedings{yang-etal-2023-local,
title = "Local Interpretation of Transformer Based on Linear Decomposition",
author = "Yang, Sen and
Huang, Shujian and
Zou, Wei and
Zhang, Jianbing and
Dai, Xinyu and
Chen, Jiajun",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.572",
doi = "10.18653/v1/2023.acl-long.572",
pages = "10270--10287",
abstract = "In recent years, deep neural networks (DNNs) have achieved state-of-the-art performance on a wide range of tasks. However, limitations in interpretability have hindered their applications in the real world. This work proposes to interpret neural networks by linear decomposition and finds that the ReLU-activated Transformer can be considered as a linear model on a single input. We further leverage the linearity of the model and propose a linear decomposition of the model output to generate local explanations. Our evaluation of sentiment classification and machine translation shows that our method achieves competitive performance in efficiency and fidelity of explanation. In addition, we demonstrate the potential of our approach in applications with examples of error analysis on multiple tasks.",
}
| In recent years, deep neural networks (DNNs) have achieved state-of-the-art performance on a wide range of tasks. However, limitations in interpretability have hindered their applications in the real world. This work proposes to interpret neural networks by linear decomposition and finds that the ReLU-activated Transformer can be considered as a linear model on a single input. We further leverage the linearity of the model and propose a linear decomposition of the model output to generate local explanations. Our evaluation of sentiment classification and machine translation shows that our method achieves competitive performance in efficiency and fidelity of explanation. In addition, we demonstrate the potential of our approach in applications with examples of error analysis on multiple tasks. | [
"Yang, Sen",
"Huang, Shujian",
"Zou, Wei",
"Zhang, Jianbing",
"Dai, Xinyu",
"Chen, Jiajun"
] | Local Interpretation of Transformer Based on Linear Decomposition | acl-long.572 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.573.bib | https://aclanthology.org/2023.acl-long.573/ | @inproceedings{viswanathan-etal-2023-datafinder,
title = "{D}ata{F}inder: Scientific Dataset Recommendation from Natural Language Descriptions",
author = "Viswanathan, Vijay and
Gao, Luyu and
Wu, Tongshuang and
Liu, Pengfei and
Neubig, Graham",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.573",
doi = "10.18653/v1/2023.acl-long.573",
pages = "10288--10303",
abstract = "Modern machine learning relies on datasets to develop and validate research ideas. Given the growth of publicly available data, finding the right dataset to use is increasingly difficult. Any research question imposes explicit and implicit constraints on how well a given dataset will enable researchers to answer this question, such as dataset size, modality, and domain. We operationalize the task of recommending datasets given a short natural language description of a research idea, to help people find relevant datasets for their needs. Dataset recommendation poses unique challenges as an information retrieval problem; datasets are hard to directly index for search and there are no corpora readily available for this task. To facilitate this task, we build the DataFinder Dataset which consists of a larger automatically-constructed training set (17.5K queries) and a smaller expert-annotated evaluation set (392 queries). Using this data, we compare various information retrieval algorithms on our test set and present a superior bi-encoder retriever for text-based dataset recommendation. This system, trained on the DataFinder Dataset, finds more relevant search results than existing third-party dataset search engines. To encourage progress on dataset recommendation, we release our dataset and models to the public.",
}
| Modern machine learning relies on datasets to develop and validate research ideas. Given the growth of publicly available data, finding the right dataset to use is increasingly difficult. Any research question imposes explicit and implicit constraints on how well a given dataset will enable researchers to answer this question, such as dataset size, modality, and domain. We operationalize the task of recommending datasets given a short natural language description of a research idea, to help people find relevant datasets for their needs. Dataset recommendation poses unique challenges as an information retrieval problem; datasets are hard to directly index for search and there are no corpora readily available for this task. To facilitate this task, we build the DataFinder Dataset which consists of a larger automatically-constructed training set (17.5K queries) and a smaller expert-annotated evaluation set (392 queries). Using this data, we compare various information retrieval algorithms on our test set and present a superior bi-encoder retriever for text-based dataset recommendation. This system, trained on the DataFinder Dataset, finds more relevant search results than existing third-party dataset search engines. To encourage progress on dataset recommendation, we release our dataset and models to the public. | [
"Viswanathan, Vijay",
"Gao, Luyu",
"Wu, Tongshuang",
"Liu, Pengfei",
"Neubig, Graham"
] | DataFinder: Scientific Dataset Recommendation from Natural Language Descriptions | acl-long.573 | Poster | 2305.16636 | [
"https://github.com/viswavi/datafinder"
] | https://huggingface.co/papers/2305.16636 | 1 | 0 | 0 | 5 | 1 | [
"viswavi/datafinder-scibert-nl-queries"
] | [] | [] |
https://aclanthology.org/2023.acl-long.574.bib | https://aclanthology.org/2023.acl-long.574/ | @inproceedings{borenstein-etal-2023-multilingual,
title = "Multilingual Event Extraction from Historical Newspaper Adverts",
author = "Borenstein, Nadav and
da Silva Perez, Nat{\'a}lia and
Augenstein, Isabelle",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.574",
doi = "10.18653/v1/2023.acl-long.574",
pages = "10304--10325",
abstract = "NLP methods can aid historians in analyzing textual materials in greater volumes than manually feasible. Developing such methods poses substantial challenges though. First, acquiring large, annotated historical datasets is difficult, as only domain experts can reliably label them. Second, most available off-the-shelf NLP models are trained on modern language texts, rendering them significantly less effective when applied to historical corpora. This is particularly problematic for less well studied tasks, and for languages other than English. This paper addresses these challenges while focusing on the under-explored task of event extraction from a novel domain of historical texts. We introduce a new multilingual dataset in English, French, and Dutch composed of newspaper ads from the early modern colonial period reporting on enslaved people who liberated themselves from enslavement. We find that: 1) even with scarce annotated data, it is possible to achieve surprisingly good results by formulating the problem as an extractive QA task and leveraging existing datasets and models for modern languages; and 2) cross-lingual low-resource learning for historical languages is highly challenging, and machine translation of the historical datasets to the considered target languages is, in practice, often the best-performing solution.",
}
| NLP methods can aid historians in analyzing textual materials in greater volumes than manually feasible. Developing such methods poses substantial challenges though. First, acquiring large, annotated historical datasets is difficult, as only domain experts can reliably label them. Second, most available off-the-shelf NLP models are trained on modern language texts, rendering them significantly less effective when applied to historical corpora. This is particularly problematic for less well studied tasks, and for languages other than English. This paper addresses these challenges while focusing on the under-explored task of event extraction from a novel domain of historical texts. We introduce a new multilingual dataset in English, French, and Dutch composed of newspaper ads from the early modern colonial period reporting on enslaved people who liberated themselves from enslavement. We find that: 1) even with scarce annotated data, it is possible to achieve surprisingly good results by formulating the problem as an extractive QA task and leveraging existing datasets and models for modern languages; and 2) cross-lingual low-resource learning for historical languages is highly challenging, and machine translation of the historical datasets to the considered target languages is, in practice, often the best-performing solution. | [
"Borenstein, Nadav",
"da Silva Perez, Nat{\\'a}lia",
"Augenstein, Isabelle"
] | Multilingual Event Extraction from Historical Newspaper Adverts | acl-long.574 | Poster | 2305.10928 | [
"https://github.com/nadavborenstein/ee-from-historical-ads"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.575.bib | https://aclanthology.org/2023.acl-long.575/ | @inproceedings{lei-etal-2023-bic,
title = "{BIC}: {T}witter Bot Detection with Text-Graph Interaction and Semantic Consistency",
author = "Lei, Zhenyu and
Wan, Herun and
Zhang, Wenqian and
Feng, Shangbin and
Chen, Zilong and
Li, Jundong and
Zheng, Qinghua and
Luo, Minnan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.575",
doi = "10.18653/v1/2023.acl-long.575",
pages = "10326--10340",
abstract = "Twitter bots are automatic programs operated by malicious actors to manipulate public opinion and spread misinformation. Research efforts have been made to automatically identify bots based on texts and networks on social media. Existing methods only leverage texts or networks alone, and while few works explored the shallow combination of the two modalities, we hypothesize that the interaction and information exchange between texts and graphs could be crucial for holistically evaluating bot activities on social media. In addition, according to a recent survey (Cresci, 2020), Twitter bots are constantly evolving while advanced bots steal genuine users{'} tweets and dilute their malicious content to evade detection. This results in greater inconsistency across the timeline of novel Twitter bots, which warrants more attention. In light of these challenges, we propose BIC, a Twitter Bot detection framework with text-graph Interaction and semantic Consistency. Specifically, in addition to separately modeling the two modalities on social media, BIC employs a text-graph interaction module to enable information exchange across modalities in the learning process. In addition, given the stealing behavior of novel Twitter bots, BIC proposes to model semantic consistency in tweets based on attention weights while using it to augment the decision process. Extensive experiments demonstrate that BIC consistently outperforms state-of-the-art baselines on two widely adopted datasets. Further analyses reveal that text-graph interactions and modeling semantic consistency are essential improvements and help combat bot evolution.",
}
| Twitter bots are automatic programs operated by malicious actors to manipulate public opinion and spread misinformation. Research efforts have been made to automatically identify bots based on texts and networks on social media. Existing methods only leverage texts or networks alone, and while few works explored the shallow combination of the two modalities, we hypothesize that the interaction and information exchange between texts and graphs could be crucial for holistically evaluating bot activities on social media. In addition, according to a recent survey (Cresci, 2020), Twitter bots are constantly evolving while advanced bots steal genuine users{'} tweets and dilute their malicious content to evade detection. This results in greater inconsistency across the timeline of novel Twitter bots, which warrants more attention. In light of these challenges, we propose BIC, a Twitter Bot detection framework with text-graph Interaction and semantic Consistency. Specifically, in addition to separately modeling the two modalities on social media, BIC employs a text-graph interaction module to enable information exchange across modalities in the learning process. In addition, given the stealing behavior of novel Twitter bots, BIC proposes to model semantic consistency in tweets based on attention weights while using it to augment the decision process. Extensive experiments demonstrate that BIC consistently outperforms state-of-the-art baselines on two widely adopted datasets. Further analyses reveal that text-graph interactions and modeling semantic consistency are essential improvements and help combat bot evolution. | [
"Lei, Zhenyu",
"Wan, Herun",
"Zhang, Wenqian",
"Feng, Shangbin",
"Chen, Zilong",
"Li, Jundong",
"Zheng, Qinghua",
"Luo, Minnan"
] | BIC: Twitter Bot Detection with Text-Graph Interaction and Semantic Consistency | acl-long.575 | Poster | 2208.08320 | [
"https://github.com/LzyFischer/BIC"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.576.bib | https://aclanthology.org/2023.acl-long.576/ | @inproceedings{patidar-etal-2023-knowledge,
title = "Do {I} have the Knowledge to Answer? Investigating Answerability of Knowledge Base Questions",
author = "Patidar, Mayur and
Faldu, Prayushi and
Singh, Avinash and
Vig, Lovekesh and
Bhattacharya, Indrajit and
{Mausam}",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.576",
doi = "10.18653/v1/2023.acl-long.576",
pages = "10341--10357",
abstract = "When answering natural language questions over knowledge bases, missing facts, incomplete schema and limited scope naturally lead to many questions being unanswerable. While answerability has been explored in other QA settings, it has not been studied for QA over knowledge bases (KBQA). We create GrailQAbility, a new benchmark KBQA dataset with unanswerability, by first identifying various forms of KB incompleteness that make questions unanswerable, and then systematically adapting GrailQA (a popular KBQA dataset with only answerable questions). Experimenting with three state-of-the-art KBQA models, we find that all three models suffer a drop in performance even after suitable adaptation for unanswerable questions. In addition, these often detect unanswerability for wrong reasons and find specific forms of unanswerability particularly difficult to handle. This underscores the need for further research in making KBQA systems robust to unanswerability.",
}
| When answering natural language questions over knowledge bases, missing facts, incomplete schema and limited scope naturally lead to many questions being unanswerable. While answerability has been explored in other QA settings, it has not been studied for QA over knowledge bases (KBQA). We create GrailQAbility, a new benchmark KBQA dataset with unanswerability, by first identifying various forms of KB incompleteness that make questions unanswerable, and then systematically adapting GrailQA (a popular KBQA dataset with only answerable questions). Experimenting with three state-of-the-art KBQA models, we find that all three models suffer a drop in performance even after suitable adaptation for unanswerable questions. In addition, these often detect unanswerability for wrong reasons and find specific forms of unanswerability particularly difficult to handle. This underscores the need for further research in making KBQA systems robust to unanswerability. | [
"Patidar, Mayur",
"Faldu, Prayushi",
"Singh, Avinash",
"Vig, Lovekesh",
"Bhattacharya, Indrajit",
"{Mausam}"
] | Do I have the Knowledge to Answer? Investigating Answerability of Knowledge Base Questions | acl-long.576 | Poster | 2212.10189 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.577.bib | https://aclanthology.org/2023.acl-long.577/ | @inproceedings{li-etal-2023-understanding,
title = "Understanding Client Reactions in Online Mental Health Counseling",
author = "Li, Anqi and
Ma, Lizhi and
Mei, Yaling and
He, Hongliang and
Zhang, Shuai and
Qiu, Huachuan and
Lan, Zhenzhong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.577",
doi = "10.18653/v1/2023.acl-long.577",
pages = "10358--10376",
abstract = "Communication success relies heavily on reading participants{'} reactions. Such feedback is especially important for mental health counselors, who must carefully consider the client{'}s progress and adjust their approach accordingly. However, previous NLP research on counseling has mainly focused on studying counselors{'} intervention strategies rather than their clients{'} reactions to the intervention. This work aims to fill this gap by developing a theoretically grounded annotation framework that encompasses counselors{'} strategies and client reaction behaviors. The framework has been tested against a large-scale, high-quality text-based counseling dataset we collected over the past two years from an online welfare counseling platform. Our study show how clients react to counselors{'} strategies, how such reactions affect the final counseling outcomes, and how counselors can adjust their strategies in response to these reactions. We also demonstrate that this study can help counselors automatically predict their clients{'} states.",
}
| Communication success relies heavily on reading participants{'} reactions. Such feedback is especially important for mental health counselors, who must carefully consider the client{'}s progress and adjust their approach accordingly. However, previous NLP research on counseling has mainly focused on studying counselors{'} intervention strategies rather than their clients{'} reactions to the intervention. This work aims to fill this gap by developing a theoretically grounded annotation framework that encompasses counselors{'} strategies and client reaction behaviors. The framework has been tested against a large-scale, high-quality text-based counseling dataset we collected over the past two years from an online welfare counseling platform. Our study show how clients react to counselors{'} strategies, how such reactions affect the final counseling outcomes, and how counselors can adjust their strategies in response to these reactions. We also demonstrate that this study can help counselors automatically predict their clients{'} states. | [
"Li, Anqi",
"Ma, Lizhi",
"Mei, Yaling",
"He, Hongliang",
"Zhang, Shuai",
"Qiu, Huachuan",
"Lan, Zhenzhong"
] | Understanding Client Reactions in Online Mental Health Counseling | acl-long.577 | Oral | 2306.15334 | [
"https://github.com/dll-wu/client-reactions"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.578.bib | https://aclanthology.org/2023.acl-long.578/ | @inproceedings{chen-etal-2023-nonlinear,
title = "Nonlinear Structural Equation Model Guided {G}aussian Mixture Hierarchical Topic Modeling",
author = "Chen, HeGang and
Mao, Pengbo and
Lu, Yuyin and
Rao, Yanghui",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.578",
doi = "10.18653/v1/2023.acl-long.578",
pages = "10377--10390",
abstract = "Hierarchical topic models, which can extract semantically meaningful topics from a textcorpus in an unsupervised manner and automatically organise them into a topic hierarchy, have been widely used to discover the underlying semantic structure of documents. However, the existing models often assume in the prior that the topic hierarchy is a tree structure, ignoring symmetrical dependenciesbetween topics at the same level. Moreover, the sparsity of text data often complicate the analysis. To address these issues, we propose NSEM-GMHTM as a deep topic model, witha Gaussian mixture prior distribution to improve the model{'}s ability to adapt to sparse data, which explicitly models hierarchical and symmetric relations between topics through the dependency matrices and nonlinear structural equations. Experiments on widely used datasets show that our NSEM-GMHTM generates more coherent topics and a more rational topic structure when compared to state-of-theart baselines. Our code is available at https: //github.com/nbnbhwyy/NSEM-GMHTM.",
}
| Hierarchical topic models, which can extract semantically meaningful topics from a textcorpus in an unsupervised manner and automatically organise them into a topic hierarchy, have been widely used to discover the underlying semantic structure of documents. However, the existing models often assume in the prior that the topic hierarchy is a tree structure, ignoring symmetrical dependenciesbetween topics at the same level. Moreover, the sparsity of text data often complicate the analysis. To address these issues, we propose NSEM-GMHTM as a deep topic model, witha Gaussian mixture prior distribution to improve the model{'}s ability to adapt to sparse data, which explicitly models hierarchical and symmetric relations between topics through the dependency matrices and nonlinear structural equations. Experiments on widely used datasets show that our NSEM-GMHTM generates more coherent topics and a more rational topic structure when compared to state-of-theart baselines. Our code is available at https: //github.com/nbnbhwyy/NSEM-GMHTM. | [
"Chen, HeGang",
"Mao, Pengbo",
"Lu, Yuyin",
"Rao, Yanghui"
] | Nonlinear Structural Equation Model Guided Gaussian Mixture Hierarchical Topic Modeling | acl-long.578 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.579.bib | https://aclanthology.org/2023.acl-long.579/ | @inproceedings{zhong-etal-2023-revisiting,
title = "Revisiting Token Dropping Strategy in Efficient {BERT} Pretraining",
author = "Zhong, Qihuang and
Ding, Liang and
Liu, Juhua and
Liu, Xuebo and
Zhang, Min and
Du, Bo and
Tao, Dacheng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.579",
doi = "10.18653/v1/2023.acl-long.579",
pages = "10391--10405",
abstract = "Token dropping is a recently-proposed strategy to speed up the pretraining of masked language models, such as BERT, by skipping the computation of a subset of the input tokens at several middle layers. It can effectively reduce the training time without degrading much performance on downstream tasks. However, we empirically find that token dropping is prone to a semantic loss problem and falls short in handling semantic-intense tasks. Motivated by this, we propose a simple yet effective semantic-consistent learning method (ScTD) to improve the token dropping. ScTD aims to encourage the model to learn how to preserve the semantic information in the representation space. Extensive experiments on 12 tasks show that, with the help of our ScTD, token dropping can achieve consistent and significant performance gains across all task types and model sizes. More encouragingly, ScTD saves up to 57{\%} of pretraining time and brings up to +1.56{\%} average improvement over the vanilla token dropping.",
}
| Token dropping is a recently-proposed strategy to speed up the pretraining of masked language models, such as BERT, by skipping the computation of a subset of the input tokens at several middle layers. It can effectively reduce the training time without degrading much performance on downstream tasks. However, we empirically find that token dropping is prone to a semantic loss problem and falls short in handling semantic-intense tasks. Motivated by this, we propose a simple yet effective semantic-consistent learning method (ScTD) to improve the token dropping. ScTD aims to encourage the model to learn how to preserve the semantic information in the representation space. Extensive experiments on 12 tasks show that, with the help of our ScTD, token dropping can achieve consistent and significant performance gains across all task types and model sizes. More encouragingly, ScTD saves up to 57{\%} of pretraining time and brings up to +1.56{\%} average improvement over the vanilla token dropping. | [
"Zhong, Qihuang",
"Ding, Liang",
"Liu, Juhua",
"Liu, Xuebo",
"Zhang, Min",
"Du, Bo",
"Tao, Dacheng"
] | Revisiting Token Dropping Strategy in Efficient BERT Pretraining | acl-long.579 | Poster | 2305.15273 | [
"https://github.com/whu-zqh/sctd"
] | https://huggingface.co/papers/2305.15273 | 1 | 0 | 0 | 7 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.580.bib | https://aclanthology.org/2023.acl-long.580/ | @inproceedings{gera-etal-2023-benefits,
title = "The Benefits of Bad Advice: Autocontrastive Decoding across Model Layers",
author = "Gera, Ariel and
Friedman, Roni and
Arviv, Ofir and
Gunasekara, Chulaka and
Sznajder, Benjamin and
Slonim, Noam and
Shnarch, Eyal",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.580",
doi = "10.18653/v1/2023.acl-long.580",
pages = "10406--10420",
abstract = "Applying language models to natural language processing tasks typically relies on the representations in the final model layer, as intermediate hidden layer representations are presumed to be less informative. In this work, we argue that due to the gradual improvement across model layers, additional information can be gleaned from the contrast between higher and lower layers during inference. Specifically, in choosing between the probable next token predictions of a generative model, the predictions of lower layers can be used to highlight which candidates are best avoided. We propose a novel approach that utilizes the contrast between layers to improve text generation outputs, and show that it mitigates degenerative behaviors of the model in open-ended generation, significantly improving the quality of generated texts. Furthermore, our results indicate that contrasting between model layers at inference time can yield substantial benefits to certain aspects of general language model capabilities, more effectively extracting knowledge during inference from a given set of model parameters.",
}
| Applying language models to natural language processing tasks typically relies on the representations in the final model layer, as intermediate hidden layer representations are presumed to be less informative. In this work, we argue that due to the gradual improvement across model layers, additional information can be gleaned from the contrast between higher and lower layers during inference. Specifically, in choosing between the probable next token predictions of a generative model, the predictions of lower layers can be used to highlight which candidates are best avoided. We propose a novel approach that utilizes the contrast between layers to improve text generation outputs, and show that it mitigates degenerative behaviors of the model in open-ended generation, significantly improving the quality of generated texts. Furthermore, our results indicate that contrasting between model layers at inference time can yield substantial benefits to certain aspects of general language model capabilities, more effectively extracting knowledge during inference from a given set of model parameters. | [
"Gera, Ariel",
"Friedman, Roni",
"Arviv, Ofir",
"Gunasekara, Chulaka",
"Sznajder, Benjamin",
"Slonim, Noam",
"Shnarch, Eyal"
] | The Benefits of Bad Advice: Autocontrastive Decoding across Model Layers | acl-long.580 | Poster | 2305.01628 | [
"https://github.com/ibm/auto-contrastive-generation"
] | https://huggingface.co/papers/2305.01628 | 1 | 0 | 0 | 7 | 1 | [
"ibm/gpt2-medium-multiexit",
"ibm/gpt-neo-125m-multiexit"
] | [] | [
"Cryptoaffiliate187/ibm-gpt-neo-125m-multiexit"
] |
https://aclanthology.org/2023.acl-long.581.bib | https://aclanthology.org/2023.acl-long.581/ | @inproceedings{rani-etal-2023-factify,
title = "{FACTIFY}-5{WQA}: 5{W} Aspect-based Fact Verification through Question Answering",
author = "Rani, Anku and
Tonmoy, S.M Towhidul Islam and
Dalal, Dwip and
Gautam, Shreya and
Chakraborty, Megha and
Chadha, Aman and
Sheth, Amit and
Das, Amitava",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.581",
doi = "10.18653/v1/2023.acl-long.581",
pages = "10421--10440",
abstract = "Automatic fact verification has received significant attention recently. Contemporary automatic fact-checking systems focus on estimating truthfulness using numerical scores which are not human-interpretable. A human fact-checker generally follows several logical steps to verify a verisimilitude claim and conclude whether it{'}s truthful or a mere masquerade. Popular fact-checking websites follow a common structure for fact categorization such as half true, half false, false, pants on fire, etc. Therefore, it is necessary to have an aspect-based (delineating which part(s) are true and which are false) explainable system that can assist human fact-checkers in asking relevant questions related to a fact, which can then be validated separately to reach a final verdict. In this paper, we propose a 5W framework (who, what, when, where, and why) for question-answer-based fact explainability. To that end, we present a semi-automatically generated dataset called FACTIFY-5WQA, which consists of 391, 041 facts along with relevant 5W QAs {--} underscoring our major contribution to this paper. A semantic role labeling system has been utilized to locate 5Ws, which generates QA pairs for claims using a masked language model. Finally, we report a baseline QA system to automatically locate those answers from evidence documents, which can serve as a baseline for future research in the field. Lastly, we propose a robust fact verification system that takes paraphrased claims and automatically validates them. The dataset and the baseline model are available at https: //github.com/ankuranii/acl-5W-QA",
}
| Automatic fact verification has received significant attention recently. Contemporary automatic fact-checking systems focus on estimating truthfulness using numerical scores which are not human-interpretable. A human fact-checker generally follows several logical steps to verify a verisimilitude claim and conclude whether it{'}s truthful or a mere masquerade. Popular fact-checking websites follow a common structure for fact categorization such as half true, half false, false, pants on fire, etc. Therefore, it is necessary to have an aspect-based (delineating which part(s) are true and which are false) explainable system that can assist human fact-checkers in asking relevant questions related to a fact, which can then be validated separately to reach a final verdict. In this paper, we propose a 5W framework (who, what, when, where, and why) for question-answer-based fact explainability. To that end, we present a semi-automatically generated dataset called FACTIFY-5WQA, which consists of 391, 041 facts along with relevant 5W QAs {--} underscoring our major contribution to this paper. A semantic role labeling system has been utilized to locate 5Ws, which generates QA pairs for claims using a masked language model. Finally, we report a baseline QA system to automatically locate those answers from evidence documents, which can serve as a baseline for future research in the field. Lastly, we propose a robust fact verification system that takes paraphrased claims and automatically validates them. The dataset and the baseline model are available at https: //github.com/ankuranii/acl-5W-QA | [
"Rani, Anku",
"Tonmoy, S.M Towhidul Islam",
"Dalal, Dwip",
"Gautam, Shreya",
"Chakraborty, Megha",
"Chadha, Aman",
"Sheth, Amit",
"Das, Amitava"
] | FACTIFY-5WQA: 5W Aspect-based Fact Verification through Question Answering | acl-long.581 | Poster | 2305.04329 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.582.bib | https://aclanthology.org/2023.acl-long.582/ | @inproceedings{mhaske-etal-2023-naamapadam,
title = "Naamapadam: A Large-Scale Named Entity Annotated Data for {I}ndic Languages",
author = "Mhaske, Arnav and
Kedia, Harshit and
Doddapaneni, Sumanth and
Khapra, Mitesh M. and
Kumar, Pratyush and
Murthy, Rudra and
Kunchukuttan, Anoop",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.582",
doi = "10.18653/v1/2023.acl-long.582",
pages = "10441--10456",
abstract = "We present, \textit{Naamapadam}, the largest publicly available Named Entity Recognition (NER) dataset for the 11 major Indian languages from two language families. The dataset contains more than 400k sentences annotated with a total of at least 100k entities from three standard entity categories (Person, Location, and, Organization) for 9 out of the 11 languages. The training dataset has been automatically created from the Samanantar parallel corpus by projecting automatically tagged entities from an English sentence to the corresponding Indian language translation. We also create manually annotated testsets for 9 languages. We demonstrate the utility of the obtained dataset on the Naamapadam-test dataset. We also release \textit{IndicNER}, a multilingual IndicBERT model fine-tuned on Naamapadam training set. IndicNER achieves an F1 score of more than 80 for 7 out of 9 test languages. The dataset and models are available under open-source licences at \url{https://ai4bharat.iitm.ac.in/naamapadam}.",
}
| We present, \textit{Naamapadam}, the largest publicly available Named Entity Recognition (NER) dataset for the 11 major Indian languages from two language families. The dataset contains more than 400k sentences annotated with a total of at least 100k entities from three standard entity categories (Person, Location, and, Organization) for 9 out of the 11 languages. The training dataset has been automatically created from the Samanantar parallel corpus by projecting automatically tagged entities from an English sentence to the corresponding Indian language translation. We also create manually annotated testsets for 9 languages. We demonstrate the utility of the obtained dataset on the Naamapadam-test dataset. We also release \textit{IndicNER}, a multilingual IndicBERT model fine-tuned on Naamapadam training set. IndicNER achieves an F1 score of more than 80 for 7 out of 9 test languages. The dataset and models are available under open-source licences at \url{https://ai4bharat.iitm.ac.in/naamapadam}. | [
"Mhaske, Arnav",
"Kedia, Harshit",
"Doddapaneni, Sumanth",
"Khapra, Mitesh M.",
"Kumar, Pratyush",
"Murthy, Rudra",
"Kunchukuttan, Anoop"
] | Naamapadam: A Large-Scale Named Entity Annotated Data for Indic Languages | acl-long.582 | Poster | 2212.10168 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.583.bib | https://aclanthology.org/2023.acl-long.583/ | @inproceedings{yu-etal-2023-crepe,
title = "{CREPE}: Open-Domain Question Answering with False Presuppositions",
author = "Yu, Xinyan and
Min, Sewon and
Zettlemoyer, Luke and
Hajishirzi, Hannaneh",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.583",
doi = "10.18653/v1/2023.acl-long.583",
pages = "10457--10480",
abstract = "When asking about unfamiliar topics, information seeking users often pose questions with false presuppositions. Most existing question answering (QA) datasets, in contrast, assume all questions have well defined answers. We introduce CREPE, a QA dataset containing a natural distribution of presupposition failures from online information-seeking forums. We find that 25{\%} of questions contain false presuppositions, and provide annotations for these presuppositions and their corrections. Through extensive baseline experiments, we show that adaptations of existing open-domain QA models can find presuppositions moderately well, but struggle when predicting whether a presupposition is factually correct. This is in large part due to difficulty in retrieving relevant evidence passages from a large text corpus. CREPE provides a benchmark to study question answering in the wild, and our analyses provide avenues for future work in better modeling and further studying the task.",
}
| When asking about unfamiliar topics, information seeking users often pose questions with false presuppositions. Most existing question answering (QA) datasets, in contrast, assume all questions have well defined answers. We introduce CREPE, a QA dataset containing a natural distribution of presupposition failures from online information-seeking forums. We find that 25{\%} of questions contain false presuppositions, and provide annotations for these presuppositions and their corrections. Through extensive baseline experiments, we show that adaptations of existing open-domain QA models can find presuppositions moderately well, but struggle when predicting whether a presupposition is factually correct. This is in large part due to difficulty in retrieving relevant evidence passages from a large text corpus. CREPE provides a benchmark to study question answering in the wild, and our analyses provide avenues for future work in better modeling and further studying the task. | [
"Yu, Xinyan",
"Min, Sewon",
"Zettlemoyer, Luke",
"Hajishirzi, Hannaneh"
] | CREPE: Open-Domain Question Answering with False Presuppositions | acl-long.583 | Oral | 2211.17257 | [
"https://github.com/velocitycavalry/crepe"
] | https://huggingface.co/papers/2211.17257 | 0 | 0 | 0 | 4 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.584.bib | https://aclanthology.org/2023.acl-long.584/ | @inproceedings{wan-etal-2023-joint,
title = "Joint Document-Level Event Extraction via Token-Token Bidirectional Event Completed Graph",
author = "Wan, Qizhi and
Wan, Changxuan and
Xiao, Keli and
Liu, Dexi and
Li, Chenliang and
Zheng, Bolong and
Liu, Xiping and
Hu, Rong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.584",
doi = "10.18653/v1/2023.acl-long.584",
pages = "10481--10492",
abstract = "We solve the challenging document-level event extraction problem by proposing a joint exaction methodology that can avoid inefficiency and error propagation issues in classic pipeline methods. Essentially, we address the three crucial limitations in existing studies. First, the autoregressive strategy of path expansion heavily relies on the orders of argument role. Second, the number of events in documents must be specified in advance. Last, unexpected errors usually exist when decoding events based on the entity-entity adjacency matrix. To address these issues, this paper designs a Token-Token Bidirectional Event Completed Graph (TT-BECG) in which the relation eType-Role1-Role2 serves as the edge type, precisely revealing which tokens play argument roles in an event of a specific event type. Exploiting the token-token adjacency matrix of the TT-BECG, we develop an edge-enhanced joint document-level event extraction model. Guided by the target token-token adjacency matrix, the predicted token-token adjacency matrix can be obtained during the model training. Then, extracted events and event records in a document are decoded based on the predicted matrix, including the graph structure and edge type decoding. Extensive experiments are conducted on two public datasets, and the results confirm the effectiveness of our method and its superiority over the state-of-the-art baselines.",
}
| We solve the challenging document-level event extraction problem by proposing a joint exaction methodology that can avoid inefficiency and error propagation issues in classic pipeline methods. Essentially, we address the three crucial limitations in existing studies. First, the autoregressive strategy of path expansion heavily relies on the orders of argument role. Second, the number of events in documents must be specified in advance. Last, unexpected errors usually exist when decoding events based on the entity-entity adjacency matrix. To address these issues, this paper designs a Token-Token Bidirectional Event Completed Graph (TT-BECG) in which the relation eType-Role1-Role2 serves as the edge type, precisely revealing which tokens play argument roles in an event of a specific event type. Exploiting the token-token adjacency matrix of the TT-BECG, we develop an edge-enhanced joint document-level event extraction model. Guided by the target token-token adjacency matrix, the predicted token-token adjacency matrix can be obtained during the model training. Then, extracted events and event records in a document are decoded based on the predicted matrix, including the graph structure and edge type decoding. Extensive experiments are conducted on two public datasets, and the results confirm the effectiveness of our method and its superiority over the state-of-the-art baselines. | [
"Wan, Qizhi",
"Wan, Changxuan",
"Xiao, Keli",
"Liu, Dexi",
"Li, Chenliang",
"Zheng, Bolong",
"Liu, Xiping",
"Hu, Rong"
] | Joint Document-Level Event Extraction via Token-Token Bidirectional Event Completed Graph | acl-long.584 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.585.bib | https://aclanthology.org/2023.acl-long.585/ | @inproceedings{zheng-etal-2023-robust,
title = "Robust Representation Learning with Reliable Pseudo-labels Generation via Self-Adaptive Optimal Transport for Short Text Clustering",
author = "Zheng, Xiaolin and
Hu, Mengling and
Liu, Weiming and
Chen, Chaochao and
Liao, Xinting",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.585",
doi = "10.18653/v1/2023.acl-long.585",
pages = "10493--10507",
abstract = "Short text clustering is challenging since it takes imbalanced and noisy data as inputs. Existing approaches cannot solve this problem well, since (1) they are prone to obtain degenerate solutions especially on heavy imbalanced datasets, and (2) they are vulnerable to noises. To tackle the above issues, we propose a Robust Short Text Clustering (RSTC) model to improve robustness against imbalanced and noisy data. RSTC includes two modules, i.e., pseudo-label generation module and robust representation learning module. The former generates pseudo-labels to provide supervision for the later, which contributes to more robust representations and correctly separated clusters. To provide robustness against the imbalance in data, we propose self-adaptive optimal transport in the pseudo-label generation module. To improve robustness against the noise in data, we further introduce both class-wise and instance-wise contrastive learning in the robust representation learning module. Our empirical studies on eight short text clustering datasets demonstrate that RSTC significantly outperforms the state-of-the-art models.",
}
| Short text clustering is challenging since it takes imbalanced and noisy data as inputs. Existing approaches cannot solve this problem well, since (1) they are prone to obtain degenerate solutions especially on heavy imbalanced datasets, and (2) they are vulnerable to noises. To tackle the above issues, we propose a Robust Short Text Clustering (RSTC) model to improve robustness against imbalanced and noisy data. RSTC includes two modules, i.e., pseudo-label generation module and robust representation learning module. The former generates pseudo-labels to provide supervision for the later, which contributes to more robust representations and correctly separated clusters. To provide robustness against the imbalance in data, we propose self-adaptive optimal transport in the pseudo-label generation module. To improve robustness against the noise in data, we further introduce both class-wise and instance-wise contrastive learning in the robust representation learning module. Our empirical studies on eight short text clustering datasets demonstrate that RSTC significantly outperforms the state-of-the-art models. | [
"Zheng, Xiaolin",
"Hu, Mengling",
"Liu, Weiming",
"Chen, Chaochao",
"Liao, Xinting"
] | Robust Representation Learning with Reliable Pseudo-labels Generation via Self-Adaptive Optimal Transport for Short Text Clustering | acl-long.585 | Poster | 2305.16335 | [
"https://github.com/hmllmh/rstc"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.586.bib | https://aclanthology.org/2023.acl-long.586/ | @inproceedings{tang-etal-2023-multilingual,
title = "Multilingual Knowledge Graph Completion with Language-Sensitive Multi-Graph Attention",
author = "Tang, Rongchuan and
Zhao, Yang and
Zong, Chengqing and
Zhou, Yu",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.586",
doi = "10.18653/v1/2023.acl-long.586",
pages = "10508--10519",
abstract = "Multilingual Knowledge Graph Completion (KGC) aims to predict missing links with multilingual knowledge graphs. However, existing approaches suffer from two main drawbacks: (a) alignment dependency: the multilingual KGC is always realized with joint entity or relation alignment, which introduces additional alignment models and increases the complexity of the whole framework; (b) training inefficiency: the trained model will only be used for the completion of one target KG, although the data from all KGs are used simultaneously. To address these drawbacks, we propose a novel multilingual KGC framework with language-sensitive multi-graph attention such that the missing links on all given KGs can be inferred by a universal knowledge completion model. Specifically, we first build a relational graph neural network by sharing the embeddings of aligned nodes to transfer language-independent knowledge. Meanwhile, a language-sensitive multi-graph attention (LSMGA) is proposed to deal with the information inconsistency among different KGs. Experimental results show that our model achieves significant improvements on the DBP-5L and E-PKG datasets.",
}
| Multilingual Knowledge Graph Completion (KGC) aims to predict missing links with multilingual knowledge graphs. However, existing approaches suffer from two main drawbacks: (a) alignment dependency: the multilingual KGC is always realized with joint entity or relation alignment, which introduces additional alignment models and increases the complexity of the whole framework; (b) training inefficiency: the trained model will only be used for the completion of one target KG, although the data from all KGs are used simultaneously. To address these drawbacks, we propose a novel multilingual KGC framework with language-sensitive multi-graph attention such that the missing links on all given KGs can be inferred by a universal knowledge completion model. Specifically, we first build a relational graph neural network by sharing the embeddings of aligned nodes to transfer language-independent knowledge. Meanwhile, a language-sensitive multi-graph attention (LSMGA) is proposed to deal with the information inconsistency among different KGs. Experimental results show that our model achieves significant improvements on the DBP-5L and E-PKG datasets. | [
"Tang, Rongchuan",
"Zhao, Yang",
"Zong, Chengqing",
"Zhou, Yu"
] | Multilingual Knowledge Graph Completion with Language-Sensitive Multi-Graph Attention | acl-long.586 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.587.bib | https://aclanthology.org/2023.acl-long.587/ | @inproceedings{adams-etal-2023-desired,
title = "What are the Desired Characteristics of Calibration Sets? Identifying Correlates on Long Form Scientific Summarization",
author = "Adams, Griffin and
Nguyen, Bichlien and
Smith, Jake and
Xia, Yingce and
Xie, Shufang and
Ostropolets, Anna and
Deb, Budhaditya and
Chen, Yuan-Jyue and
Naumann, Tristan and
Elhadad, No{\'e}mie",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.587",
doi = "10.18653/v1/2023.acl-long.587",
pages = "10520--10542",
abstract = "Summarization models often generate text that is poorly calibrated to quality metrics because they are trained to maximize the likelihood of a single reference (MLE). To address this, recent work has added a calibration step, which exposes a model to its own ranked outputs to improve relevance or, in a separate line of work, contrasts positive and negative sets to improve faithfulness. While effective, much of this work has focused on \textit{how} to generate and optimize these sets. Less is known about \textit{why} one setup is more effective than another. In this work, we uncover the underlying characteristics of effective sets. For each training instance, we form a large, diverse pool of candidates and systematically vary the subsets used for calibration fine-tuning. Each selection strategy targets distinct aspects of the sets, such as lexical diversity or the size of the gap between positive and negatives. On three diverse scientific long-form summarization datasets (spanning biomedical, clinical, and chemical domains), we find, among others, that faithfulness calibration is optimal when the negative sets are extractive and more likely to be generated, whereas for relevance calibration, the metric margin between candidates should be maximized and surprise{--}the disagreement between model and metric defined candidate rankings{--}minimized.",
}
| Summarization models often generate text that is poorly calibrated to quality metrics because they are trained to maximize the likelihood of a single reference (MLE). To address this, recent work has added a calibration step, which exposes a model to its own ranked outputs to improve relevance or, in a separate line of work, contrasts positive and negative sets to improve faithfulness. While effective, much of this work has focused on \textit{how} to generate and optimize these sets. Less is known about \textit{why} one setup is more effective than another. In this work, we uncover the underlying characteristics of effective sets. For each training instance, we form a large, diverse pool of candidates and systematically vary the subsets used for calibration fine-tuning. Each selection strategy targets distinct aspects of the sets, such as lexical diversity or the size of the gap between positive and negatives. On three diverse scientific long-form summarization datasets (spanning biomedical, clinical, and chemical domains), we find, among others, that faithfulness calibration is optimal when the negative sets are extractive and more likely to be generated, whereas for relevance calibration, the metric margin between candidates should be maximized and surprise{--}the disagreement between model and metric defined candidate rankings{--}minimized. | [
"Adams, Griffin",
"Nguyen, Bichlien",
"Smith, Jake",
"Xia, Yingce",
"Xie, Shufang",
"Ostropolets, Anna",
"Deb, Budhaditya",
"Chen, Yuan-Jyue",
"Naumann, Tristan",
"Elhadad, No{\\'e}mie"
] | What are the Desired Characteristics of Calibration Sets? Identifying Correlates on Long Form Scientific Summarization | acl-long.587 | Oral | 2305.07615 | [
"https://github.com/griff4692/calibrating-summaries"
] | https://huggingface.co/papers/2305.07615 | 3 | 1 | 1 | 10 | 1 | [] | [
"griffin/ChemSum"
] | [] |
https://aclanthology.org/2023.acl-long.588.bib | https://aclanthology.org/2023.acl-long.588/ | @inproceedings{gandhi-etal-2023-annotating,
title = "Annotating Mentions Alone Enables Efficient Domain Adaptation for Coreference Resolution",
author = "Gandhi, Nupoor and
Field, Anjalie and
Strubell, Emma",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.588",
doi = "10.18653/v1/2023.acl-long.588",
pages = "10543--10558",
abstract = "Although recent neural models for coreference resolution have led to substantial improvements on benchmark datasets, it remains a challenge to successfully transfer these models to new target domains containing many out-of-vocabulary spans and requiring differing annotation schemes. Typical approaches involve continued training on annotated target-domain data, but obtaining annotations is costly and time-consuming. In this work, we show that adapting mention detection is the key component to successful domain adaptation of coreference models, rather than antecedent linking. We also show annotating mentions alone is nearly twice as fast as annotating full coreference chains. Based on these insights, we propose a method for efficiently adapting coreference models, which includes a high-precision mention detection objective and requires only mention annotations in the target domain. Extensive evaluation across three English coreference datasets: CoNLL-2012 (news/conversation), i2b2/VA (medical notes), and child welfare notes, reveals that our approach facilitates annotation-efficient transfer and results in a 7-14{\%} improvement in average F1 without increasing annotator time.",
}
| Although recent neural models for coreference resolution have led to substantial improvements on benchmark datasets, it remains a challenge to successfully transfer these models to new target domains containing many out-of-vocabulary spans and requiring differing annotation schemes. Typical approaches involve continued training on annotated target-domain data, but obtaining annotations is costly and time-consuming. In this work, we show that adapting mention detection is the key component to successful domain adaptation of coreference models, rather than antecedent linking. We also show annotating mentions alone is nearly twice as fast as annotating full coreference chains. Based on these insights, we propose a method for efficiently adapting coreference models, which includes a high-precision mention detection objective and requires only mention annotations in the target domain. Extensive evaluation across three English coreference datasets: CoNLL-2012 (news/conversation), i2b2/VA (medical notes), and child welfare notes, reveals that our approach facilitates annotation-efficient transfer and results in a 7-14{\%} improvement in average F1 without increasing annotator time. | [
"G",
"hi, Nupoor",
"Field, Anjalie",
"Strubell, Emma"
] | Annotating Mentions Alone Enables Efficient Domain Adaptation for Coreference Resolution | acl-long.588 | Oral | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.589.bib | https://aclanthology.org/2023.acl-long.589/ | @inproceedings{xu-etal-2023-universal,
title = "A Universal Discriminator for Zero-Shot Generalization",
author = "Xu, Haike and
Lin, Zongyu and
Zhou, Jing and
Zheng, Yanan and
Yang, Zhilin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.589",
doi = "10.18653/v1/2023.acl-long.589",
pages = "10559--10575",
abstract = "Generative modeling has been the dominant approach for large-scale pretraining and zero-shot generalization. In this work, we challenge this convention by showing that discriminative approaches perform substantially better than generative ones on a large number of NLP tasks. Technically, we train a single discriminator to predict whether a text sample comes from the true data distribution, similar to GANs. Since many NLP tasks can be formulated as selecting from a few options, we use this discriminator to predict the concatenation of input and which option has the highest probability of coming from the true data distribution. This simple formulation achieves state-of-the-art zero-shot results on the T0 benchmark, outperforming T0 by 16.0{\%}, 7.8{\%}, and 11.5{\%} respectively on different scales. In the finetuning setting, our approach also achieves new state-of-the-art results on a wide range of NLP tasks, with only 1/4 parameters of previous methods. Meanwhile, our approach requires minimal prompting efforts, which largely improves robustness and is essential for real-world applications. Furthermore, we also jointly train a generalized UD in combination with generative tasks, which maintains its advantage on discriminative tasks and simultaneously works on generative tasks.",
}
| Generative modeling has been the dominant approach for large-scale pretraining and zero-shot generalization. In this work, we challenge this convention by showing that discriminative approaches perform substantially better than generative ones on a large number of NLP tasks. Technically, we train a single discriminator to predict whether a text sample comes from the true data distribution, similar to GANs. Since many NLP tasks can be formulated as selecting from a few options, we use this discriminator to predict the concatenation of input and which option has the highest probability of coming from the true data distribution. This simple formulation achieves state-of-the-art zero-shot results on the T0 benchmark, outperforming T0 by 16.0{\%}, 7.8{\%}, and 11.5{\%} respectively on different scales. In the finetuning setting, our approach also achieves new state-of-the-art results on a wide range of NLP tasks, with only 1/4 parameters of previous methods. Meanwhile, our approach requires minimal prompting efforts, which largely improves robustness and is essential for real-world applications. Furthermore, we also jointly train a generalized UD in combination with generative tasks, which maintains its advantage on discriminative tasks and simultaneously works on generative tasks. | [
"Xu, Haike",
"Lin, Zongyu",
"Zhou, Jing",
"Zheng, Yanan",
"Yang, Zhilin"
] | A Universal Discriminator for Zero-Shot Generalization | acl-long.589 | Poster | 2211.08099 | [
"https://github.com/rafa-zy/ud"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.590.bib | https://aclanthology.org/2023.acl-long.590/ | @inproceedings{bailly-etal-2023-syntax,
title = "Syntax and Geometry of Information",
author = {Bailly, Rapha{\"e}l and
Leblond, Laurent and
G{\'a}bor, Kata},
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.590",
doi = "10.18653/v1/2023.acl-long.590",
pages = "10576--10595",
abstract = "This paper presents an information-theoretical model of syntactic generalization. We study syntactic generalization from the perspective of the capacity to disentangle semantic and structural information, emulating the human capacity to assign a grammaticality judgment to semantically nonsensical sentences. In order to isolate the structure, we propose to represent the probability distribution behind a corpus as the product of the probability of a semantic context and the probability of a structure, the latter being independent of the former. We further elaborate the notion of abstraction as a relaxation of the property of independence. It is based on the measure of structural and contextual information for a given representation. We test abstraction as an optimization objective on the task of inducing syntactic categories from natural language data and show that it significantly outperforms alternative methods. Furthermore, we find that when syntax-unaware optimization objectives succeed in the task, their success is mainly due to an implicit disentanglement process rather than to the model structure. On the other hand, syntactic categories can be deduced in a principled way from the independence between structure and context.",
}
| This paper presents an information-theoretical model of syntactic generalization. We study syntactic generalization from the perspective of the capacity to disentangle semantic and structural information, emulating the human capacity to assign a grammaticality judgment to semantically nonsensical sentences. In order to isolate the structure, we propose to represent the probability distribution behind a corpus as the product of the probability of a semantic context and the probability of a structure, the latter being independent of the former. We further elaborate the notion of abstraction as a relaxation of the property of independence. It is based on the measure of structural and contextual information for a given representation. We test abstraction as an optimization objective on the task of inducing syntactic categories from natural language data and show that it significantly outperforms alternative methods. Furthermore, we find that when syntax-unaware optimization objectives succeed in the task, their success is mainly due to an implicit disentanglement process rather than to the model structure. On the other hand, syntactic categories can be deduced in a principled way from the independence between structure and context. | [
"Bailly, Rapha{\\\"e}l",
"Leblond, Laurent",
"G{\\'a}bor, Kata"
] | Syntax and Geometry of Information | acl-long.590 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.591.bib | https://aclanthology.org/2023.acl-long.591/ | @inproceedings{wang-etal-2023-greenkgc,
title = "{G}reen{KGC}: A Lightweight Knowledge Graph Completion Method",
author = "Wang, Yun Cheng and
Ge, Xiou and
Wang, Bin and
Kuo, C.-C. Jay",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.591",
doi = "10.18653/v1/2023.acl-long.591",
pages = "10596--10613",
abstract = "Knowledge graph completion (KGC) aims to discover missing relationships between entities in knowledge graphs (KGs). Most prior KGC work focuses on learning embeddings for entities and relations through a simple score function. Yet, a higher-dimensional embedding space is usually required for a better reasoning capability, which leads to larger model size and hinders applicability to real-world problems (e.g., large-scale KGs or mobile/edge computing). A lightweight modularized KGC solution, called GreenKGC, is proposed in this work to address this issue. GreenKGC consists of three modules: representation learning, feature pruning, and decision learning, to extract discriminant KG features and make accurate predictions on missing relationships using classifiers and negative sampling. Experimental results demonstrate that, in low dimensions, GreenKGC can outperform SOTA methods in most datasets. In addition, low-dimensional GreenKGC can achieve competitive or even better performance against high-dimensional models with a much smaller model size.",
}
| Knowledge graph completion (KGC) aims to discover missing relationships between entities in knowledge graphs (KGs). Most prior KGC work focuses on learning embeddings for entities and relations through a simple score function. Yet, a higher-dimensional embedding space is usually required for a better reasoning capability, which leads to larger model size and hinders applicability to real-world problems (e.g., large-scale KGs or mobile/edge computing). A lightweight modularized KGC solution, called GreenKGC, is proposed in this work to address this issue. GreenKGC consists of three modules: representation learning, feature pruning, and decision learning, to extract discriminant KG features and make accurate predictions on missing relationships using classifiers and negative sampling. Experimental results demonstrate that, in low dimensions, GreenKGC can outperform SOTA methods in most datasets. In addition, low-dimensional GreenKGC can achieve competitive or even better performance against high-dimensional models with a much smaller model size. | [
"Wang, Yun Cheng",
"Ge, Xiou",
"Wang, Bin",
"Kuo, C.-C. Jay"
] | GreenKGC: A Lightweight Knowledge Graph Completion Method | acl-long.591 | Poster | 2208.09137 | [
"https://github.com/yunchengwang/greenkgc"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.592.bib | https://aclanthology.org/2023.acl-long.592/ | @inproceedings{do-etal-2023-unsupervised,
title = "Unsupervised Open-domain Keyphrase Generation",
author = "Do, Lam and
Akash, Pritom Saha and
Chang, Kevin Chen-Chuan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.592",
doi = "10.18653/v1/2023.acl-long.592",
pages = "10614--10627",
abstract = "In this work, we study the problem of unsupervised open-domain keyphrase generation, where the objective is a keyphrase generation model that can be built without using human-labeled data and can perform consistently across domains. To solve this problem, we propose a seq2seq model that consists of two modules, namely phraseness and informativeness module, both of which can be built in an unsupervised and open-domain fashion. The phraseness module generates phrases, while the informativeness module guides the generation towards those that represent the core concepts of the text. We thoroughly evaluate our proposed method using eight benchmark datasets from different domains. Results on in-domain datasets show that our approach achieves state-of-the-art results compared with existing unsupervised models, and overall narrows the gap between supervised and unsupervised methods down to about 16{\%}. Furthermore, we demonstrate that our model performs consistently across domains, as it surpasses the baselines on out-of-domain datasets.",
}
| In this work, we study the problem of unsupervised open-domain keyphrase generation, where the objective is a keyphrase generation model that can be built without using human-labeled data and can perform consistently across domains. To solve this problem, we propose a seq2seq model that consists of two modules, namely phraseness and informativeness module, both of which can be built in an unsupervised and open-domain fashion. The phraseness module generates phrases, while the informativeness module guides the generation towards those that represent the core concepts of the text. We thoroughly evaluate our proposed method using eight benchmark datasets from different domains. Results on in-domain datasets show that our approach achieves state-of-the-art results compared with existing unsupervised models, and overall narrows the gap between supervised and unsupervised methods down to about 16{\%}. Furthermore, we demonstrate that our model performs consistently across domains, as it surpasses the baselines on out-of-domain datasets. | [
"Do, Lam",
"Akash, Pritom Saha",
"Chang, Kevin Chen-Chuan"
] | Unsupervised Open-domain Keyphrase Generation | acl-long.592 | Poster | 2306.10755 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.593.bib | https://aclanthology.org/2023.acl-long.593/ | @inproceedings{jiang-etal-2023-cognitive,
title = "A Cognitive Stimulation Dialogue System with Multi-source Knowledge Fusion for Elders with Cognitive Impairment",
author = "Jiang, Jiyue and
Wang, Sheng and
Li, Qintong and
Kong, Lingpeng and
Wu, Chuan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.593",
doi = "10.18653/v1/2023.acl-long.593",
pages = "10628--10640",
abstract = "When communicating with elders with cognitive impairment, cognitive stimulation (CS) help to maintain the cognitive health of elders. Data sparsity is the main challenge in building CS-based dialogue systems, particularly in the Chinese language. To fill this gap, we construct a Chinese CS conversation (CSConv) dataset, which contains about 2.6K groups of dialogues with therapy principles and emotional support strategy labels. Making chit chat while providing emotional support is overlooked by the majority of existing cognitive dialogue systems. In this paper, we propose a multi-source knowledge fusion method for CS dialogue (CSD), to generate open-ended responses guided by the therapy principle and emotional support strategy. We first use a progressive mask method based on external knowledge to learn encoders as effective classifiers, which is the prerequisite to predict the therapy principle and emotional support strategy of the target response. Then a decoder interacts with the perceived therapy principle and emotional support strategy to generate responses. Extensive experiments conducted on the CSConv dataset demonstrate the effectiveness of the proposed method, while there is still a large space for improvement compared to human performance.",
}
| When communicating with elders with cognitive impairment, cognitive stimulation (CS) help to maintain the cognitive health of elders. Data sparsity is the main challenge in building CS-based dialogue systems, particularly in the Chinese language. To fill this gap, we construct a Chinese CS conversation (CSConv) dataset, which contains about 2.6K groups of dialogues with therapy principles and emotional support strategy labels. Making chit chat while providing emotional support is overlooked by the majority of existing cognitive dialogue systems. In this paper, we propose a multi-source knowledge fusion method for CS dialogue (CSD), to generate open-ended responses guided by the therapy principle and emotional support strategy. We first use a progressive mask method based on external knowledge to learn encoders as effective classifiers, which is the prerequisite to predict the therapy principle and emotional support strategy of the target response. Then a decoder interacts with the perceived therapy principle and emotional support strategy to generate responses. Extensive experiments conducted on the CSConv dataset demonstrate the effectiveness of the proposed method, while there is still a large space for improvement compared to human performance. | [
"Jiang, Jiyue",
"Wang, Sheng",
"Li, Qintong",
"Kong, Lingpeng",
"Wu, Chuan"
] | A Cognitive Stimulation Dialogue System with Multi-source Knowledge Fusion for Elders with Cognitive Impairment | acl-long.593 | Poster | 2305.08200 | [
""
] | https://huggingface.co/papers/2305.08200 | 0 | 0 | 0 | 5 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-long.594.bib | https://aclanthology.org/2023.acl-long.594/ | @inproceedings{zhang-etal-2023-plug,
title = "Plug-and-Play Knowledge Injection for Pre-trained Language Models",
author = "Zhang, Zhengyan and
Zeng, Zhiyuan and
Lin, Yankai and
Wang, Huadong and
Ye, Deming and
Xiao, Chaojun and
Han, Xu and
Liu, Zhiyuan and
Li, Peng and
Sun, Maosong and
Zhou, Jie",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.594",
doi = "10.18653/v1/2023.acl-long.594",
pages = "10641--10658",
abstract = "Injecting external knowledge can improve the performance of pre-trained language models (PLMs) on various downstream NLP tasks. However, massive retraining is required to deploy new knowledge injection methods or knowledge bases for downstream tasks. In this work, we are the first to study how to improve the flexibility and efficiency of knowledge injection by reusing existing downstream models. To this end, we explore a new paradigm \textit{plug-and-play knowledge injection}, where knowledge bases are injected into frozen existing downstream models by a \textit{knowledge plugin}. Correspondingly, we propose a plug-and-play injection method \textit{map-tuning}, which trains a mapping of knowledge embeddings to enrich model inputs with mapped embeddings while keeping model parameters frozen. Experimental results on three knowledge-driven NLP tasks show that existing injection methods are not suitable for the new paradigm, while map-tuning effectively improves the performance of downstream models. Moreover, we show that a frozen downstream model can be well adapted to different domains with different mapping networks of domain knowledge. Our code and models are available at \url{https://github.com/THUNLP/Knowledge-Plugin}.",
}
| Injecting external knowledge can improve the performance of pre-trained language models (PLMs) on various downstream NLP tasks. However, massive retraining is required to deploy new knowledge injection methods or knowledge bases for downstream tasks. In this work, we are the first to study how to improve the flexibility and efficiency of knowledge injection by reusing existing downstream models. To this end, we explore a new paradigm \textit{plug-and-play knowledge injection}, where knowledge bases are injected into frozen existing downstream models by a \textit{knowledge plugin}. Correspondingly, we propose a plug-and-play injection method \textit{map-tuning}, which trains a mapping of knowledge embeddings to enrich model inputs with mapped embeddings while keeping model parameters frozen. Experimental results on three knowledge-driven NLP tasks show that existing injection methods are not suitable for the new paradigm, while map-tuning effectively improves the performance of downstream models. Moreover, we show that a frozen downstream model can be well adapted to different domains with different mapping networks of domain knowledge. Our code and models are available at \url{https://github.com/THUNLP/Knowledge-Plugin}. | [
"Zhang, Zhengyan",
"Zeng, Zhiyuan",
"Lin, Yankai",
"Wang, Huadong",
"Ye, Deming",
"Xiao, Chaojun",
"Han, Xu",
"Liu, Zhiyuan",
"Li, Peng",
"Sun, Maosong",
"Zhou, Jie"
] | Plug-and-Play Knowledge Injection for Pre-trained Language Models | acl-long.594 | Poster | 2305.17691 | [
"https://github.com/thunlp/knowledge-plugin"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.595.bib | https://aclanthology.org/2023.acl-long.595/ | @inproceedings{zhou-etal-2023-two,
title = "Two Birds One Stone: Dynamic Ensemble for {OOD} Intent Classification",
author = "Zhou, Yunhua and
Yang, Jianqiang and
Wang, Pengyu and
Qiu, Xipeng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.595",
doi = "10.18653/v1/2023.acl-long.595",
pages = "10659--10673",
abstract = "Out-of-domain (OOD) intent classification is an active field of natural language understanding, which is of great practical significance for intelligent devices such as the Task-Oriented Dialogue System. It mainly contains two challenges: it requires the model to know what it knows and what it does not know. This paper investigates {``}overthinking{''} in the open-world scenario and its impact on OOD intent classification. Inspired by this, we propose a two-birds-one-stone method, which allows the model to decide whether to make a decision on OOD classification early during inference and can ensure accuracy and accelerate inference. At the same time, to adapt to the behavior of dynamic inference, we also propose a training method based on ensemble methods. In addition to bringing certain theoretical insights, we also conduct detailed experiments on three real-world intent datasets. Compared with the previous baselines, our method can not only improve inference speed, but also achieve significant performance improvements.",
}
| Out-of-domain (OOD) intent classification is an active field of natural language understanding, which is of great practical significance for intelligent devices such as the Task-Oriented Dialogue System. It mainly contains two challenges: it requires the model to know what it knows and what it does not know. This paper investigates {``}overthinking{''} in the open-world scenario and its impact on OOD intent classification. Inspired by this, we propose a two-birds-one-stone method, which allows the model to decide whether to make a decision on OOD classification early during inference and can ensure accuracy and accelerate inference. At the same time, to adapt to the behavior of dynamic inference, we also propose a training method based on ensemble methods. In addition to bringing certain theoretical insights, we also conduct detailed experiments on three real-world intent datasets. Compared with the previous baselines, our method can not only improve inference speed, but also achieve significant performance improvements. | [
"Zhou, Yunhua",
"Yang, Jianqiang",
"Wang, Pengyu",
"Qiu, Xipeng"
] | Two Birds One Stone: Dynamic Ensemble for OOD Intent Classification | acl-long.595 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.596.bib | https://aclanthology.org/2023.acl-long.596/ | @inproceedings{laban-etal-2023-swipe,
title = "{SW}i{PE}: A Dataset for Document-Level Simplification of {W}ikipedia Pages",
author = "Laban, Philippe and
Vig, Jesse and
Kryscinski, Wojciech and
Joty, Shafiq and
Xiong, Caiming and
Wu, Chien-Sheng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.596",
doi = "10.18653/v1/2023.acl-long.596",
pages = "10674--10695",
abstract = "Text simplification research has mostly focused on sentence-level simplification, even though many desirable edits - such as adding relevant background information or reordering content - may require document-level context. Prior work has also predominantly framed simplification as a single-step, input-to-output task, only implicitly modeling the fine-grained, span-level edits that elucidate the simplification process. To address both gaps, we introduce the SWiPE dataset, which reconstructs the document-level editing process from English Wikipedia (EW) articles to paired Simple Wikipedia (SEW) articles. In contrast to prior work, SWiPE leverages the entire revision history when pairing pages in order to better identify simplification edits. We work with Wikipedia editors to annotate 5,000 EW-SEW document pairs, labeling more than 40,000 edits with proposed 19 categories. To scale our efforts, we propose several models to automatically label edits, achieving an F-1 score of up to 70.9, indicating that this is a tractable but challenging NLU task. Finally, we categorize the edits produced by several simplification models and find that SWiPE-trained models generate more complex edits while reducing unwanted edits.",
}
| Text simplification research has mostly focused on sentence-level simplification, even though many desirable edits - such as adding relevant background information or reordering content - may require document-level context. Prior work has also predominantly framed simplification as a single-step, input-to-output task, only implicitly modeling the fine-grained, span-level edits that elucidate the simplification process. To address both gaps, we introduce the SWiPE dataset, which reconstructs the document-level editing process from English Wikipedia (EW) articles to paired Simple Wikipedia (SEW) articles. In contrast to prior work, SWiPE leverages the entire revision history when pairing pages in order to better identify simplification edits. We work with Wikipedia editors to annotate 5,000 EW-SEW document pairs, labeling more than 40,000 edits with proposed 19 categories. To scale our efforts, we propose several models to automatically label edits, achieving an F-1 score of up to 70.9, indicating that this is a tractable but challenging NLU task. Finally, we categorize the edits produced by several simplification models and find that SWiPE-trained models generate more complex edits while reducing unwanted edits. | [
"Laban, Philippe",
"Vig, Jesse",
"Kryscinski, Wojciech",
"Joty, Shafiq",
"Xiong, Caiming",
"Wu, Chien-Sheng"
] | SWiPE: A Dataset for Document-Level Simplification of Wikipedia Pages | acl-long.596 | Poster | 2305.19204 | [
"https://github.com/salesforce/simplification"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.597.bib | https://aclanthology.org/2023.acl-long.597/ | @inproceedings{li-etal-2023-message,
title = "Are Message Passing Neural Networks Really Helpful for Knowledge Graph Completion?",
author = "Li, Juanhui and
Shomer, Harry and
Ding, Jiayuan and
Wang, Yiqi and
Ma, Yao and
Shah, Neil and
Tang, Jiliang and
Yin, Dawei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.597",
doi = "10.18653/v1/2023.acl-long.597",
pages = "10696--10711",
abstract = "Knowledge graphs (KGs) facilitate a wide variety of applications. Despite great efforts in creation and maintenance, even the largest KGs are far from complete. Hence, KG completion (KGC) has become one of the most crucial tasks for KG research. Recently, considerable literature in this space has centered around the use of Message Passing (Graph) Neural Networks (MPNNs), to learn powerful embeddings. The success of these methods is naturally attributed to the use of MPNNs over simpler multi-layer perceptron (MLP) models, given their additional message passing (MP) component. In this work, we find that surprisingly, simple MLP models are able to achieve comparable performance to MPNNs, suggesting that MP may not be as crucial as previously believed. With further exploration, we show careful scoring function and loss function design has a much stronger influence on KGC model performance. This suggests a conflation of scoring function design, loss function design, and MP in prior work, with promising insights regarding the scalability of state-of-the-art KGC methods today, as well as careful attention to more suitable MP designs for KGC tasks tomorrow.",
}
| Knowledge graphs (KGs) facilitate a wide variety of applications. Despite great efforts in creation and maintenance, even the largest KGs are far from complete. Hence, KG completion (KGC) has become one of the most crucial tasks for KG research. Recently, considerable literature in this space has centered around the use of Message Passing (Graph) Neural Networks (MPNNs), to learn powerful embeddings. The success of these methods is naturally attributed to the use of MPNNs over simpler multi-layer perceptron (MLP) models, given their additional message passing (MP) component. In this work, we find that surprisingly, simple MLP models are able to achieve comparable performance to MPNNs, suggesting that MP may not be as crucial as previously believed. With further exploration, we show careful scoring function and loss function design has a much stronger influence on KGC model performance. This suggests a conflation of scoring function design, loss function design, and MP in prior work, with promising insights regarding the scalability of state-of-the-art KGC methods today, as well as careful attention to more suitable MP designs for KGC tasks tomorrow. | [
"Li, Juanhui",
"Shomer, Harry",
"Ding, Jiayuan",
"Wang, Yiqi",
"Ma, Yao",
"Shah, Neil",
"Tang, Jiliang",
"Yin, Dawei"
] | Are Message Passing Neural Networks Really Helpful for Knowledge Graph Completion? | acl-long.597 | Poster | 2205.10652 | [
"https://github.com/Juanhui28/Are_MPNNs_helpful"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.598.bib | https://aclanthology.org/2023.acl-long.598/ | @inproceedings{corro-2023-dynamic,
title = "A dynamic programming algorithm for span-based nested named-entity recognition in $O(n^2)$",
author = "Corro, Caio",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.598",
doi = "10.18653/v1/2023.acl-long.598",
pages = "10712--10724",
abstract = "Span-based nested named-entity recognition (NER) has a cubic-time complexity using avariant of the CYK algorithm. We show that by adding a supplementary structural constraint on the search space, nested NER has a quadratic-time complexity, that is the same asymptotic complexity than the non-nested case. The proposed algorithm covers a large part of three standard English benchmarks and delivers comparable experimental results.",
}
| Span-based nested named-entity recognition (NER) has a cubic-time complexity using avariant of the CYK algorithm. We show that by adding a supplementary structural constraint on the search space, nested NER has a quadratic-time complexity, that is the same asymptotic complexity than the non-nested case. The proposed algorithm covers a large part of three standard English benchmarks and delivers comparable experimental results. | [
"Corro, Caio"
] | A dynamic programming algorithm for span-based nested named-entity recognition in O(n^2) | acl-long.598 | Poster | 2210.04738 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.599.bib | https://aclanthology.org/2023.acl-long.599/ | @inproceedings{bao-etal-2023-target,
title = "Target-Side Augmentation for Document-Level Machine Translation",
author = "Bao, Guangsheng and
Teng, Zhiyang and
Zhang, Yue",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.599",
doi = "10.18653/v1/2023.acl-long.599",
pages = "10725--10742",
abstract = "Document-level machine translation faces the challenge of data sparsity due to its long input length and a small amount of training data, increasing the risk of learning spurious patterns. To address this challenge, we propose a target-side augmentation method, introducing a data augmentation (DA) model to generate many potential translations for each source document. Learning on these wider range translations, an MT model can learn a smoothed distribution, thereby reducing the risk of data sparsity. We demonstrate that the DA model, which estimates the posterior distribution, largely improves the MT performance, outperforming the previous best system by 2.30 s-BLEU on News and achieving new state-of-the-art on News and Europarl benchmarks.",
}
| Document-level machine translation faces the challenge of data sparsity due to its long input length and a small amount of training data, increasing the risk of learning spurious patterns. To address this challenge, we propose a target-side augmentation method, introducing a data augmentation (DA) model to generate many potential translations for each source document. Learning on these wider range translations, an MT model can learn a smoothed distribution, thereby reducing the risk of data sparsity. We demonstrate that the DA model, which estimates the posterior distribution, largely improves the MT performance, outperforming the previous best system by 2.30 s-BLEU on News and achieving new state-of-the-art on News and Europarl benchmarks. | [
"Bao, Guangsheng",
"Teng, Zhiyang",
"Zhang, Yue"
] | Target-Side Augmentation for Document-Level Machine Translation | acl-long.599 | Poster | 2305.04505 | [
"https://github.com/baoguangsheng/target-side-augmentation"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
Subsets and Splits